|Home | About | Journals | Submit | Contact Us | Français|
Single cell quantitative real-time PCR (qRT-PCR) combined with high-throughput arrays allows analysis of gene expression profiles at a molecular level in approximately 11 hours after cell sample collection. We present here a high-content microfluidic real-time platform as a powerful tool to comparatively investigate regulation of developmental processes in single cells. This approach overcomes the limitations involving heterogeneous cell populations and sample amounts, and may shed light on differential regulation of gene expression in normal versus disease-related contexts. Furthermore, high-throughput single-cell qRT-PCR provides a standardized, comparative assay for in-depth analysis of the mechanisms underlying human pluripotent stem cell self-renewal and differentiation.
Novel genomic technologies have paved the way for a more detailed understanding of fundamental cell biology processes in the past three decades. However, this experimental background is mainly based on analysing pools of several thousand cells, which limits in-depth analysis of cell differentiation, disease mechanisms, and the regulatory function underlying the transcriptome. Single cell quantitative real-time PCR (single cell qRT-PCR), a rapidly evolving tool, can address individual cellular units1–7. Minority cell populations are particularly important for disease-relevant questions such as stem cell differentiation and cancer, and single cell qRT-PCR proves to be especially useful in these contexts. In general, single cells are small and compartmentalized units characterized by numerous transcriptional variations and an increase of gene expression noise3, 8–10. The functional redundancy between genes of any cell type may account for varying gene expression levels and differences in the biological outcome, respectively. At the same time, variability may be averaged out to guarantee sufficient expression levels of critical genes at all time. Likely, these processes are subject to tight functional control by dynamic and reversible regulatory mechanisms1. While several approaches have been developed to analyse heterogeneity in single cells11–13, a more comprehensive understanding of these issues requires high-throughput approaches that can dissect molecular mechanisms in a homogeneous system of single-cell populations11. Microfluidic chips are a powerful approach to comparatively measure and characterize whole transcriptome patterns at high resolution7, 14.
Recently, microfluidic real-time arrays have been employed for gene expression profiling of human induced pluripotent stem cells (hiPSCs)15, providing a new platform to quantitatively evaluate the features of hiPSCs by analyzing the expression of a large pluripotency and differentiation markers panel, based on single cells16. Reprogramming of somatic cells to an induced pluripotent state using defined factors offers significant advantages over application of human embryonic stem cells (hESCs)17. Although they are considered the “gold standard”, hESCs are also compromised by ethical concerns and limited availability. By contrast, patient-derived hiPSCs avoid these pitfalls and therefore hold great hopes for drug discovery, genetic correction, and regenerative medicine purposes. However, before application of hiPSCs can be safely applied in human patients, the mechanisms behind hiPSCs’ pluripotency, self-renewal, and differentiation need to be further explored.
Here we describe in detail a microfluidic platform for evaluation of gene expression profiles in single pluripotent cells15 thereby providing vital information on variations in pluripotency within a defined population of cells, such as an hiPSC colony. Therefore, this technique will allow in-depth analysis of pluripotent cell fate, also with regard to monitoring the proliferation and survival of hiPSCs in animal models via in vivo imaging, as only a relatively small numbers of cells need to be extracted and characterized. hiPSCs and hESCs gene expression profiles can be directly compared, as performed recently employing microfluidic real-time arrays15. In this direct evaluation of hiPSC versus hESC, significant heterogeneity was seen at the single cell level for hiPSCs, which may account for their less consistent cardiac and endothelial cell differentiation efficiency as well as slower proliferation rate in vivo.
Microfluidic single cell qRTPCR reveals gene expression differences amongst different types of single cells, such as present in pluripotent stem cell colonies. Currently, researchers explore a growing range of methods for functional analysis of heterogeneity in single stem cells, such as live imaging or genetic labelling18, 19. Other available methods include microfluidic real-time PCR on chips14, 20 or microdroplet-based microfluidic technology21. The former is suitable for applications such as large-scale profiling comparing gene expression levels in different tissues or single cells, while the latter provides an alternative approach for real-time gene expression profiling in small pools of cells. On the other hand, this protocol not only covers microfluidic real-time analysis of gene expression on a single-cell basis, thereby providing a novel approach for testing large numbers of cells and genes of interest simultaneously and in a high-throughput manner. Here, we also feature the application of microfluidic single cell qRTPCR for hiPSCs and hESCs, which are subject of a novel and thriving research field. While other present applications focus on specific cell types such as bacteria7, an advantage of this protocol lies in its general and wide applicability to experimental setups even when the amounts of starting material are limiting. Generating data quality that rivals benchmark real-time qPCR results, the microfluidic single cell gene expression platform presented here surpasses the classical approach of RNA isolation from a pool of cells as it offers the unique opportunity of addressing possible cell population heterogeneity. Overall, this platform allows quantitative high-content analysis and comparison of variations in gene expression patterns at the single cell level. A potential limitation of this method is the requirement of a sufficient number of replicates to achieve statistically conclusive data on single cell level.
Yet overall, various fields including early developmental biology, systems biology, and molecular medicine will benefit from this technology, especially when limited numbers of cells are available for analysis22–25. Researchers only begin to explore the potentially vast possibilities presented by this state-of-the-art microfluidic single cell gene expression profiling to elucidate some fundamental yet complex biological processes such as cellular self-renewal and differentiation, transcriptional control, variability and redundancy at the molecular level16, 26, 27.
Single cell real-time PCR does not require RNA isolation due to the fact that the sample is a single cell, and therefore only contains picograms of RNA which would be lost during a traditional RNA isolation (Fig. 1). Instead, the sorted cell is directly introduced into the reverse transcription-specific target amplification (RT-STA) master mix. In order to achieve optimal sample preservation, this master mix contains 2x CellDirect Reaction Mix (CellsDirect One-Step qRT-PCR kit, Invitrogen) and SUPERase-In (Applied Biosystems), an enzyme which protects RNA extracted from lysed cell from any RNases that could be present in the RT-STA master mix. The master mix also presents a pool of all Taqman primers that will be studied later during the real-time PCR on the dynamic array. Finally, the master mix contains a mixture of two different enzymes for the RT-STA reaction. SuperScript III Reverse Transcriptase (Invitrogen), an improved engineered enzyme version with reduced RNase H activity and increased thermal stability, is capable of synthesizing specific cDNA from a total RNA sample. The second enzyme in the master mix, Platinum Taq DNA polymerase (Invitrogen), incorporates an antibody for inhibition of polymerase activity at room temperature (25 °C). Its automatic “hot start” allows for full polymerase activity restoration, increasing the amplification efficiency.
Each chip-run should include positive and negative RNA controls. When sorting the cells into the 96-well plate containing the RT-STA master mix, at least 4 wells can be left empty (no cell will be sorted into these wells). Two of these wells are used for positive control and they will have 1 μl positive RNA control at concentration of 0.1 μg/μl. The other two wells are left empty. The guiding principle in choosing the positive RNA control is to obtain high numbers of all assayed transcripts, such as total RNA extracted from a cell line known to express the desired genes at high levels. For instance, to represent genes associated with development, testes RNA (Human Testes Total RNA, Ambion) can be used.
Although single cell qRT-PCR is a useful tool for any type of cell, this protocol focuses on pluripotent cells, such as hESCs and hiPSCs. Both cell types are cultured on Matrigel-coated plates under conditions promoting an undifferentiated state. Despite these culturing conditions, spontaneous differentiation may occur at the edge of the colonies. In order to exclude partially differentiating cells from the analysis, hESCs and hiPSCs are fluorescently labelled using SSEA-4 (stage specific embryonic antigen-4) and Tra-1-60 pluripotency markers28, 29 as targets, as well as propidium iodide (PI) to distinguish the live cell population. When sorting cells for transcriptome studies it is important to choose suitable cell surface markers that do not alter the transcriptome profile because of changes in signalling pathways. Cells positive for Alexa Fluor 647-conjugated SSEA-4 and Alexa Fluor 488-conjugated TRA-1-60 are sorted using 488nm and 640nm lasers and 525/50 and 670/30 band pass filtering, respectively. One double-positive, live cell is directly sorted into each well of a 96-well plate, which is a highly automated process. Flow rates are set at 1000–2000 events per second. Five-laser (355nm, 405nm, 488nm, 532n and 640nm) FACSAria II provided with FACSDiva software for cell sorting is used (http://facs.stanford.edu/). The steps for FACS calibration are outlined in Box 1.
|1||To calibrate PMT voltages||Run fluorescent beads (e.g., BD Calibrite).|
|2||To set drop delay||Run Accudrop beads (e.g., BD Accudrop).|
|3||To set up compensation (if necessary)||Run single-stained cells or antibody-labeled beads.|
|4||To remove debris and doubles||Create a parent gate based on FSC-A vs SSC-A properties. Subsequently, create a daughter gate based on FSC-A vs FSC-W, and granddaughter gate based on SSC-A vs SSC-W properties.|
|5||To remove dead cells||Exclude Propidium Iodine- or 7AAD-positive events from the analysis.|
|6||To set up sorting gates||Run an aliquot of the stained cells to be sorted. Set up positive and negative gates for the single cell sort.|
In order to characterize complex cellular states at the single cell level using Dynamic Array™ integrated fluidic circuits (IFCs) several different steps must be followed for sample preparation. Prior to the qRT-PCR, reverse transcription and specific target amplification in a thermocycler suitable for 96-well plates are necessary. Both reactions occur in the same well that a particular cell was sorted into. After cells have been lysed in RT-STA buffer by influx of fluids through their cell membrane driven by hypotonic pressure, RNA is released and reversely transcribed at 50°C for 15 minutes via the SuperScript III Reverse Transcriptase. After cDNA synthesis, the reverse transcription reaction is inhibited and Platinum Taq DNA is activated at 70°C for 2 minutes. Activation of the enzyme is followed by specific target amplification for 18 cycles, each cycle consisting of a denaturation step (95°C for 15 seconds) followed by annealing/extension (60°C for 4 minutes). The specific target transcript amplification by Platinum Taq is required prior to real-time PCR via the microfluidic array because the amounts of RNA present in a single cell are on the order of picograms, and can be as low as a 1–10 molecules for rare mRNA species of interest. The Dynamic Array™ microfluidic chip operates by partitioning the sample into 48 (or 96) microfluidic chambers and performing qPCR detection and quantification for a specific gene in each chamber. The specific target amplification enriches all loci of interest such that the cDNA synthesized from those loci can later be distributed evenly throughout the 48 (or 96) Dynamic Array™ chambers, with each chamber having at least 100 copies of even the rarest target transcript. Without this first amplification round, the target genes would not be efficiently detected by Dynamic Array™ IFC.
Two Dynamic Array™ IFC sizes are available for the single cell real-time PCR in nanoliter reaction volumes: 48.48 and 96.96. In this protocol, we focus on the use of 48.48 Dynamic Array™ IFCs, which will be run in the BioMark HD Reader. With this array, the expression levels of 48 genes may be studied. Among the 48 genes there should be at least one housekeeping gene which reflects the amount of RNA supplied and is necessary for result data normalization. The suitability of housekeeping genes depends on the cell type and the experimental set up. The most abundant and stable housekeeping gene, and therefore widely used, is the ribosomal RNA 18S, which is also the one selected in the current experimental protocol. The chip is built as a matrix of 48 (or 96) “sample” channels crossing 48 (or 96) “gene expression assay” channels intersecting in 48*48=2,304 (or 96*96=9,216) chambers30. In each of these chambers, a specific sample (single cell cDNA) is combined with a specific gene expression assay. Each gene expression assay consists of a pair of primers targeting the gene of interest, in addition to a gene-specific dual labelled hydrolysis probe or a dsDNA binding dye. In each of these chambers, a single real-time qPCR reaction takes place. Dynamic Array™ IFCs offers the opportunity to combine different standard reagents, an advantage that makes the assay configuration more flexible. The protocol we detail here is intended for the use of Taqman primers (Applied Biosystems). Alternatively, DNA-binding dyes such as EvaGreen (Biotium) may be used, allowing for a more affordable selection of real-time PCR primers.
In this study, the single cell real-time PCR results are shown as threshold cycles (CT). A CT value, which measures target transcript abundance in the sample, is defined as the qPCR cycle for which the relative fluorescence intensity exceeds a common threshold within the exponential qPCR amplification phase. These values are calculated by the Fluidigm Real-time PCR analysis software. The results can be displayed as results table, image view diagram, or heat map. The results table shows the numeric CT values of the different samples for each gene. The image view option allows to graphically plot the fluorescence intensity as it increases during the qPCR amplification. Lastly, the heat map represents the results according to color range, with each color tone indicating a CT value. In heat map display, individual assays (X-axis) are plotted against individual samples (Y-axis).
The housekeeping gene included in the array can be used to normalize the CT values and thereby to correct differences in the CT values that are due to slightly varying starting amounts of RNA. The normalization of the CT values results in so-called ΔCT values is shown by the following:
ΔCT values from the sample of interest can be related to a control sample:
to obtain fold difference values following the formula:
The primary data obtained are compatible with different gene expression software.
Prepare DPBS with 2% ES-Cell FBS. Store at 4°C for ~1 week.
Dilute the Matrigel 1/60 in DMEM-F12 media. Thaw the Matrigel on ice. Prepare the dilution fresh just before coating the plates.
Add 1 ml of Matrigel solution to cover the whole surface of a 6-well-plate well. Incubate the 6-well plate for a minimum of 30 min at 37°C. Before using the well, aspirate the excess of Matrigel. Do not let the well dry. Remove the Matrigel just prior to add the media.
Mix the supplement bottle content with the mTeSR-1 media. Store at 4°C for ~2 weeks or make aliquots and store at −20°C for ~2 months.
|Component||Volume to add per reaction (μl)||Final concentration|
|20x Taqman assay 1||1.5||0.2x|
|20x Taqman assay 2||1.5||0.2x|
|20x Taqman assay 48||1.5||0.2x|
0.2x Assay mix can be stored at 4°C in the darkness for ~ 1 week
|Component||Volume to add per reaction (μl)||Final concentration|
|2x CellsDirect Reaction Mix||5||1x|
|SuperScript™ III/Platinum Taq Mix||0.2|
|50°C for 15 min|
|70°C for 2 min|
|18 cycles of: 95°C for 15 sec, 60°C for 4 min|
|Hold at 4°C|
|50°C for 2 min||Amplification Erase Phase|
|95°C for 10 min||Hot Start Phase|
|40 cycles of: 95°C for 15 sec, 60°C for 1 min||Denature phase, anneal phase|
Troubleshooting advice is summarized in Table 1.
Successful chip-runs will generate heatmaps with gene Ct values encoded by different colours (Fig 5). In an optimal situation, all the samples rows loaded with cell cDNA will show Ct values at least at the housekeeping gene column. However, depending on the experimental design, not all the gene columns may yield a value. The lack of Ct values for a specific gene, represented in the heatmap in black, can be read as the absence of gene expression in that sample. In a bad chip run the housekeeping gene levels will be barely or not detected, or the same cell type will create overly different transcriptome patterns which will hinder result interpretation.
We are grateful to Patricia E. de Almeida for FACS discussion. We acknowledge funding support from Swiss National Science Foundation PBBEP3_129803 (V.S.F.); German Research Foundation (A.D.E.); Howard Hughes Medical Institute (S.R.Q.); National Institutes of Health (NIH) DP2OD004437, RC1AG036142, R01AI085575, and Burroughs Welcome Foundation (J.C.W.).
AUTHOR CONTRIBUTIONS V.S.F. and A.D.E. prepared most of the paper. T.K., S.R.Q., and J.C.W provided advice and proofread the paper.
COMPETING FINANCIAL INTERESTS S.R.Q. is affiliated with Fluidigm. The rest of the authors declare no competing financial interests.