Real-Time PCR Technical Guide, Vol. X

Table of Contents

Letter from the Editor
Index of Experts
Q1: Which are your preferred cell-capture and lysis approaches, and why?
Q2: How do you produce single-cell cDNA libraries suitable for qPCR?
Q3: What are your tips for improving single-cell qPCR experimental design?
Q4: What steps do you take to reduce experimental error and minimize noise?
Q5: How do you quantify your single-cell qPCR results?
Q6: Which bioinformatics tools do you use for your analyses?
List of Resources

Download the PDF version here

Letter from the Editor

Editor's note: This online version has been updated from a previous version, to properly position Anders Ståhlberg's contributions to Q3-Q5. Those responses are shifted in the print and PDF versions of this guide. GT regrets the error.

In an April 2010 Trends in Biotechnology paper, Lawrence Berkeley National Laboratory's Daojing Wang and San Francisco-based consultant Steven Bodovitz dubbed single-cell analysis "the new frontier in 'omics.'" Now, single-cell genomics, transcriptomics, proteomics, and metabolomics studies are no longer out of reach. Using emerging micro- and nanofluidic approaches as well as standby molecular techniques, researchers increasingly aim to extract singlecell signals from population noise.

For the 10th installment of its real-time PCR technical guide series, Genome Technology sought expert advice on applying a singlecell focus to a trusty technique. Here, qPCR pioneers share their tips for analyzing single cells — addressing everything from cell capture and lysis to ensuring reproducible results. (False positives, it seems, are pervasive problems, though can be all-but avoided with a few tricks.)

Once you've had your real-time PCR fix in the following pages, be sure to check the handy reference guide (beginning on page 13) for a selection of recent papers of relevance, suggested by our experts.

— Tracy Vence

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.

Mikael Kubista
TATAA Biocenter

Anders Ståhlberg
University of Gothenberg

Finn-Arne Weltzien
Norwegian School of Veterinary Science

Weiwen Zhang
The Biodesign institute
Arizona State University

Editor's note: Zhang is now at Tianjin University.

Q1: Which are your preferred cell-capture and lysis approaches, and why?

Most frequently we use fluorescence-activated cell sorting. For samples that can be disintegrated and dissociated, FACS is most convenient for high-throughput work, and is readily integrated into our workflow. We also use immunomagnetic enrichment of specific cell types — for example, to collect cancer cells from circulation. For lysis, we use our own CelluLyser or Roche [Diagnostics'] Real-Time Ready Cell Lysis [kit], which has been optimized for compatibility with reverse transcription.

— Mikael Kubista

Microaspiration, flow cytometry, laser capture microdissection, or automated microdevices can be applied to collect individual cells. Many variations have been developed to fit specific applications.

Every setup has its pros and cons. At the moment, I am using flow cytometry to collect various cell types. FACS allows cell sorting into PCR plates that are compatible with any RT-qPCR instrumentation. Furthermore, FACS allows cell enrichment using specific cell markers. The throughput of FACS is significantly higher compared to that of microaspiration and laser capture microdissection. One drawback of FACS is that you cannot do a visual inspection of the cell to determine important parameters, such as cell morphology. FACS also requires a single-cell suspension; consequently, the history of the cell is lost.

To avoid any RNA losses, we use lysis buffers that are compatible with downstream enzymatic reactions without any further purification. Several lysis options are available including CelluLyser (TATAA Biocenter), Real-Time Ready Cell Lysis [kit] (Roche Diagnostics), Single Cell-to-CT (Life Technologies), and noncommercial buffers. We have noted that different cell types require different lysis buffers. We prefer to use weak lysis buffers — water and noncommercial buffers — to avoid enzymatic inhibition in downstream steps. If stronger lysis buffers are needed, I have had good experience using CelluLyser and Real-Time Ready Cell Lysis buffer.

— Anders Ståhlberg

In our experience, false positives are a major problem in single- cell qPCR, especially when working with primary dispersed cells. As a result, from primary dispersed cells, we prefer to harvest cytosol through a thin glass (patch) pipette. When using cell lines, harvesting whole cells usually works fine.

— Finn-Arne Weltzien

We have used several methods for single-cell isolation in our published research, such as dilution to extinction, cell trapping, and cell micromanipulation. The preferred method is really dependent on what the nature of the samples and the research goals are.

We have successfully used a cell micromanipulator to isolate single eukaryotic and bacterial cells under microscopes. The single cells can be captured and loaded into PCR tubes (0.2 mL) or microtube (1.5 mL) preloaded with 40 μL 1×PBS buffer. The cell can then be lysed using commercial kits (like the ZR RNA MicroPrep Kit, Zymo Research) for RNA isolation or by heat or chemical lysis for RT-PCR directly. One drawback of the micromanipulation method is that throughput is typically not high.

— Weiwen Zhang

Q2: How do you produce single-cell cDNA libraries suitable for qPCR?

We use exclusively PCR-based pre-amplification. The TaqMan PreAmp Master Mix (Life Technologies) is, in our view, the best dedicated kit, although we have excellent results with our own master mixes as well for many systems. It is most important to handle the samples properly. It is easy to go wrong without experience. We offer single cell profiling service in Europe including digital PCR measurements for accurate copy number determinations in collaboration with Life Technologies. Starting in 2012, we will offer hands-on training on single-cell expression profiling.

— Mikael Kubista

We do reverse transcription directly in the cytosol/cell lysate using cells direct and random hexamer primers.

— Finn-Arne Weltzien

After RNA isolation from single cells, a suitable reverse transcription procedure is applied. The cDNA generated is typically enough to measure a dozen highly expressed genes from single cells. However, if more genes need to be measured, the cDNA from a single cell needs to be amplified. There are several methods available for cDNA amplification in the literature, however, their potential biases still need further evaluation.

— Weiwen Zhang

Q3: What are your tips for improving single-cell qPCR experimental design?

There is a limit to how many markers can be measured per cell, which makes selection of markers critical. The strategy selecting markers depends on the purpose of the study. If the objective is to study expression pathways and gene networks, quite different selection criteria apply than if the objective is to distinguish cell types. Once criteria have been specified statistical selection tools are available in software such as GenEx.

— Mikael Kubista

Single-cell gene expression profiling usually requires many RT-qPCR experiments. Before starting any single-cell experiments, cell population measurements should be performed and analyzed to guide the design of any single-cell experiments. Cell population measurements will hopefully give you useful information about what experimental conditions to use and a rough idea about the expression level of your target genes.

Single-cell analysis is useful in combination with cell population measurements to fingerprint a few carefully selected biological conditions. Analyzing several conditions and biological replicates are usually not feasible due to cost and time limitations. For the first round of experiments, we usually measure about 10 genes in 40 cells to 100 cells, avoiding pre-amplification. This data will give you information about cell heterogeneity and the expression levels of your key markers. Then, it will be easier to design a larger experiment.

— Anders Ståhlberg

Be sure to include robust control experiments to avoid false positives. Also be extremely careful in preparing and performing experiments to avoid RNA degradation or suboptimal reverse-transcription conditions.

— Finn-Arne Weltzien

In general, the efficiency of the following three steps is important for a successful single-cell analysis: cell isolation (i.e. time needed and success rate), RNA isolation and cDNA synthesis, and primer design for single-cell qPCR. We have found that primers working well for high-abundance template don't necessary work well for qPCR template of single-cell level.

— Weiwen Zhang

Q4: What steps do you take to reduce experimental error and minimize noise?

It is very important to establish experimental steps that are reproducible. For example, lysis should be tested by dividing the lysate into aliquots that are analyzed separately and compared. Using the CelluLyser we find excellent agreement between aliquots evidencing the cell was lysed homogeneously. Using heat or distilled water only to lyse cells may produce heterogeneous lysate resulting in unequal aliquots. Recovery efficiency and reproducibility can be tested using an RNA spike. We have designed a universal spike with A-tail and 5'-cap to mimic native mRNA. The spike can be added to the lysate or microinjected into the cell. Reverse transcription should be optimized. RT yield varies up to 200- fold on priming strategy, target gene, and reverse transcriptase. Also, the amount of lysate added to the RT mix must be carefully matched to avoid inhibition, while maximizing RT yield, and the amount of cDNA added to the PCR. Pre-ampl i f icat ion shal l also be careful l y opt imized and validated. In general, reproducibility is more critical than minimizing bias, although the two parameters are typically closely correlated.

— Mikael Kubista

All experimental steps need to be carefully optimized. First make sure that your collected cells are representative for the population and have good RNA quality. Minimize all dilutions between the different experimental steps. This is important, since the total number of molecules is low. However, be sure to not inhibit the next experimental step by loading too high a concentration of detergents, reverse transcriptase, et cetera, in the following step. Reverse transcription, pre-amplification, and qPCR are all sensitive to inhibition. For example, reverse transcriptase may inhibit both pre-amplification and qPCR if too much is loaded into these reactions. The dilution factor depends on enzyme choice and brand. If few genes are analyzed, pre-amplification may be avoided to eliminate one experimental step that may introduce additional bias. You gain little by running technical replicates. Instead, collect and analyze more cells if possible.

— Anders Ståhlberg

By being extremely meticulous when preparing and performing experiments. We also coat the glass pipette with a hydrophobic substance to avoid charged nucleic acids to stick to the glass — this would easily produce false positives. Remember that the qPCR primers need to be of the highest quality to produce good quality results.

— Finn-Arne Weltzien

Errors between technical replicates are in general very small, based on our published results. However, errors between biological replicates are hard to be determined due to the fact that no perfect control exists. To minimize the errors between single cells, we have typically run all single-cell qPCR analysis (including RNA isolation, cDNA synthesis, and real time PCR) in parallel, and in a single PCR heating plate (i.e. ABI 48-well, 96-well, or 384-well format).

— Weiwen Zhang

Q5: How do you quantify your single-cell qPCR results?

Analysis of single-cell profiling data involves three steps: normalization, scaling, and clustering. Normalization with the expression of reference genes — common when analyzing classical samples — is not applicable to single cells because of the large, uncorrelated variation in genes' expressions. We showed this in our original publication, and it has since then been demonstrated for virtually all genes and all kinds of cells, the exception being cells without active mRNA metabolism such as oocytes. Better is to normalize expression per cell, essentially comparing the measured quantities directly. This is most intuitive and works very well. Of course, such an approach does not account for losses during processing of the samples, but we have found those are usually negligible. If there is concern about the processing, the universal RNA spike mentioned above can be used to measure yields and establish reproducibility. As a further control, we are exploring normalization with the expression of Alu-repeat containing transcripts, which reflects the average expression of a very large number of transcripts and may provide a stable norm. But, as said, for most systems normalization per cell is adequate.

Data are analyzed with multivariate methods for expression profiling. Data are either unscaled, mean-centered, or autoscaled. Mean centering (subtracting the average value) and autoscaling (subtracting the average value and dividing with the standard deviation) is done traditionally along genes to even out or equalize the weights of the markers in the analyses. For single cells, we have in some cases also found mean centering and autoscaling along samples useful. This is unconventional in multivariate statistics, but effectively corresponds to global normalization.

— Mikael Kubista

Data can be transformed to absolute cDNA numbers if pre-amplification has been avoided, otherwise data analysis can be performed with Cq-values, since transcript levels are log-normally distributed in individual cells. Initially, it is useful to plot data in various ways, such as heat maps and violin plots. In addition basic statistical analysis should be performed (i.e. number of positive cells, mean, and variation). Cell population measurements are usually normalized against validated reference genes. This normalizing strategy should be avoided in singlecell analysis, since all transcript levels vary over time in individual cells. We have found that correlation analysis and unsupervised algorithms, such as Kohonen self- organizing maps, are useful to define subpopulations and gene networks.

— Anders Ståhlberg

We are careful in using single-cell qPCR for quantitative measures. We usually use it only for qualitative analyses. If we quantify however, we use copy number of target gene relative to copy number of a reference gene.

— Finn-Arne Weltzien

For each single cell, we run qPCR analysis for target genes and reference gene, such as 16s rRNA for prokaryotic cells and 28s or actin for eukaryotic cells. However, since we also noticed that expression level of these commonly used reference genes (i.e. 16s, 28s rRNA) could be quite different between individual cells, caution needs to be taken with standard quantification rule for bulk cells qPCR: [relative mRNA level] A = 2-Δ(Ct A – Ct 28s). In most cases, we reported both raw and normalized Ct values and used them for quantification of our single-cell qPCR results.

— Weiwen Zhang

Q6: Which bioinformatics tools do you use for your analyses?

We use GenEx for single-cell analysis. In fact, we use GenEx for all our qPCR data analysis. GenEx is powerful, user-friendly, and it supports all leading qPCR instruments, which makes it easy to import data and annotations from experiments. GenEx has excellent data-quality assessment, which is important in single cell work, and powerful tools for single-cell expression profiling. It has an intuitive user-friendly interface that gives even the inexperienced user easy access to powerful tools such as scatter plots, hierarchical clustering, heat map, principal component analysis, self-organizing maps, potential curves, support vector machines, and correlation analysis. There will be a dedicated session on single-cell data analysis at the next qPCR symposium.

— Mikael Kubista

I perform my data analysis in GenEx and SPSS.

— Anders Ståhlberg

For sample size of a few dozen genes, ANOVA student t-test or other simple statistic tools are applied. However, if several thousand genes are analyzed using amplified single-cell cDNA as template, more advanced bioinformatics tools may be needed.

— Weiwen Zhang

List of Resources

Here, our experts point out handy resources to help answer all your real-time PCR questions.

Publications

Bengtsson M, Hemberg M, Rorsman P, Ståhlberg A. (2008). Quantification of mRNA in single cells and modeling of RT-qPCR induced noise. BMC Molecular Biology. 9(63).

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: 1388-1392.

Bustin S, Bergkvist A, Nolan T. (2011). In silico tools for qPCR assay design and data analysis. Methods in Molecular Biology. 760:283-306.

Citri A, Pang ZP, Südhof TC, Wernig M, Malenka RC. (2011). Comprehensive qPCR profiling of gene expression in single neuronal cells. Nature Protocols. 7(1):118-127.

Dalerba P, Kalisky T, Sahoo D, Rajendran PS, Rothenberg ME, Leyrat AA, Sim S, Okamoto J, Johnston DM, Qian D, Zabala M, Bueno J, Neff NF, Wang J, Shelton AA, Visser B, Hisamori S, Shimono Y, van de Wetering M, Clevers H, Clarke MF, Quake SR. (2011). Single-cell dissection of transcriptional heterogeneity in human colon tumors. Nature Biotechnology. 29(12): 1120-1127.

Forlenza M, Kaiser T, Savelkoul HF, Wiegertjes GF. (2012). The use of real-time quantitative PCR for the analysis of cytokine mRNA levels. Methods in Molecular Biology. 820: 7-23.

Gao W, Zhang W, Meldrum DR. (2011). RTqPCR based quantitative analysis of gene expression in single bacterial cells. Journal of Microbiological Methods. 85: 221-222.

Gubelmann C, Gattiker A, Massouras A, Hens K, David F, Decouttere F, Rougemont J, Deplancke B. (2011). GETPrime: a gene- or transcript-specific primer database for quantitative real-time PCR. Database: the Journal of Biological Databases and Curation. 2011:bar040.

Hannemann J, Meyer-Staeckling S, Kemming D, Alpers I, Joosse SA, Pospisil H, Kurtz S, Görndt J, Püschel K, Riethdorf S, Pantel K, Brandt B. (2011). Quantitative high-resolution genomic analysis of single cancer cells. PLoS One. 6(11):e26362.

Hodne K, Haug TM, Weltzien FA. (2011). Single-cell qPCR on dispersed primary pituitary cells — an optimized protocol. BMC Molecular Biology. 11: 82.

Pabinger S, Thallinger GG, Snajder R, Eichhorn H, Rader R, Trajanoski Z. (2009). QPCR: Application for real-time PCR data management and analysis. BMC Bioinformatics. 10: 268.

Reiter M, Kirchner B, Müller H, Holzhauer C, Mann W, Pfaffl MW. (2011). Quantification noise in single cell experiments. Nucleic Acids Research. 39(18):e124.

Ståhlberg A, Andersson D, Aurelius J, Faiz M, Pekna M, Kubista M, and Pekny M. (2011). Defining cell populations with single-cell gene expression profiling: correlations and identification of astrocyte subpopulations. Nucleic Acids Research. Epub doi: 10.1093/nar/gkq1182.

Ståhlberg A, Håkansson J, Xian X, Semb H, and Kubista M. (2004). Properties of the reverse transcription reaction in mRNA quantification. Clinical Chemistry. 50(3): 509-515.

Wang D, Bodovitz S. (2010). Single-cell analysis: the new frontier in 'omics.' Trends in Biotechnology. 28(6): 281-290.

Wang X, Spandidos A, Wang H, Seed B. (2012). PrimerBank: a PCR primer database for quantitative gene expression analysis — 2012 Update. Nucleic Acids Research. 40(1): D1144-1149.

White AK, VanInsberghe M, Petriv OI, Hamidi M, Sikorski D, Marra MA, Piret J, Aparicio S, Hansen CL. (2011). High-throughput microfluidic single-cell RT-qPCR. Proceedings of the National Academy of Sciences. 108(34): 13,999-14,004.

Zeng J, Wang J, Gao W, Mohammadreza A, Kelbauskas L, Zhang W, Johnson RH, Meldrum DR. (2011). Quantitative single-cell gene expression measurements of multiple genes in response to hypoxia treatment. Analytical and Bioanalytical Chemistry. 401: 3-13.

Conferences

TATAA Biocenter Courses (Various, including: 2- or 3-Day Hands-On PCR, Single-Cell Analysis, and 2- or 3-Day Experimental Design and Statistical Data Analysis for qPCR)
Göteborg, Sweden, and Prague, various times
(visit http://www.tataa.com/Courses/Courses.html for further information)

Genomics Research: RNAi and miRNA, Advances in qPCR, Epigenetics, and Next-Gen Sequencing
April 21, 2012
Boston, MA

Single-Cell Biology and Real-Time PCR 2012
May 1-2, 2012
Waltham, MA

EuroPCR
May 12-15, 2012
Paris, France

Genomics Research Europe: RNAi and miRNA, Advances in qPCR, Epigenetics, and Agrigenomics
September 4-5, 2012
Frankfurt, Germany

Single-Cell Analysis Summit
September 25-26, 2012
San Diego, CA

qPCR Symposium
November 13-16, 2012
DaeJeon, South Korea

Web sites

qPCR Forum
http:// http://www.qpcrforum.com/

qPCR-Related Discussions on Protocol Online
http:// http://www.protocol-online.org/biology-forums/qPCR.html

FastPCR (University of Helsinki)
http://primerdigital.com/fastpcr.html