Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
The data presented in this study are available on request from the corresponding author.
Reverse transcription quantitative PCR (RT-qPCR) has delivered significant insights in understanding the gene expression landscape. Thanks to its precision, sensitivity, flexibility, and cost effectiveness, RT-qPCR has also found utility in advanced single-cell analysis. Single-cell RT-qPCR now represents a well-established method, suitable for an efficient screening prior to single-cell RNA sequencing (scRNA-Seq) experiments, or, oppositely, for validation of hypotheses formulated from high-throughput approaches. Here, we aim to provide a comprehensive summary of the scRT-qPCR method by discussing the limitations of single-cell collection methods, describing the importance of reverse transcription, providing recommendations for the preamplification and primer design, and summarizing essential data processing steps. With the detailed protocol attached in the appendix, this tutorial provides a set of guidelines that allow any researcher to perform scRT-qPCR measurements of the highest standard.
Keywords: single cell, sample collection, reverse transcription, preamplification, quantitative PCR, gene expression, RT-qPCR
Gene expression profiling has accelerated our knowledge on the progression of many diseases and injuries [1,2,3,4,5,6,7,8]. Originally, gene expression was studied on tissue autopsies, providing sufficient amounts of starting material for most applications. The incentive to study disease, development, or healing with greater detail shifted the direction toward single-cell profiling, which allows: (i) identifying and/or discovering novel cell subtypes [4,5,9]; (ii) understanding cell trajectories undertaken upon stimuli [10,11,12]; (iii) describing the heterogeneity of tissue composition [13,14,15]; and (iv) uncovering the cell type-specific interactions within complex tissues [11,16].
Traditionally, single-cell reverse transcription quantitative PCR (scRT-qPCR) represented the first method of choice for conducting single-cell gene expression measurements. In the last decade, its position was replaced by single-cell RNA sequencing (scRNA-Seq) that allows measurement of the complete transcriptome in thousands of cells in a single experiment. Despite this trend, scRT-qPCR still retains a strong position in the field thanks to its precision, sensitivity, wide dynamic range, and ease of use [17,18,19,20,21,22,23,24,25].
Although there have been several attempts to summarize the scRT-qPCR workflow, these are either outdated publications, not reflecting the recent development in the field, or they provide only a theoretical background, or pay attention to a specific part of the workflow [22,23,26,27,28]. Here, we aim to extend the discussion about the individual aspects of single-cell experiments [29,30], summarizing recent updates in the field and providing a practical guideline for scRT-qPCR measurement, translating the theory into hands-on practice. Collectively, we reviewed individual aspects of single-cell collection and material handling, reverse transcription (RT), preamplification (preAMP), quantitative PCR (qPCR), and data analysis ( Figure 1 ). The Supplementary Material encloses a detailed exemplary protocol serving as a starting point for researchers in conducting their own scRT-qPCR measurements (Supplementary File S1).
The scRT-qPCR workflow begins with the preparation of a single-cell suspension followed by the collection of single cells ( Figure 2 ). While the former is simple for cell cultures or biofluids, dissociation of tissues may pose a serious challenge. A literature survey and on-site optimization are recommended to avoid common issues associated with inadequate protocols for preparation of single-cell suspensions, i.e., low yield, low viability, loss of vulnerable cell types, or activation of immediate-early genes [31,32,33]. The yield and viability are routinely examined using counting chambers or automated cell counters in combination with adequate staining (trypan blue, propidium iodide). The losses of vulnerable cell types may be controlled by inspection of antibody-labeled cells in the microscope or an RT-qPCR measurement of cell type marker gene expression before and after the dissociation procedure [34]. Lastly, the activation of stress-induced genes is minimized by low temperatures during dissociation and/or the use of transcriptional inhibitors [35,36]. As the use of low temperatures may hamper the dissociation efficiency, the application of psychrophilic proteases has been recently suggested [37].
Compared to scRNA-Seq, scRT-qPCR faces the challenge of a relatively low throughput. Unguided collection of cells from a heterogenous tissue is therefore non-effective, as targeted cell types may be collected only in a few cases per experiment. Therefore, cell populations of interest are typically identified by labeling delivered via gene constructs or antibodies conjugated with a fluorescence signal. The most established methods for collection of single-cell material are fluorescence-activated cell sorting (FACS), micromanipulation, and laser capture microdissection (LCM), although a few modern alternatives have been recently introduced as well [38]. The first two methods retrieve live single cells, whereas the last method usually requires material fixation. Minimizing the collection time reduces the risk of artificial alteration of the gene expression landscape [31]. Out of the three methods, only FACS can be looked upon as a quick, high-throughput method. In comparison, micromanipulation and LCM are laborious and time-consuming techniques, but they allow for visual inspection of the material. In addition, LCM retains the information on the spatial composition of the collected material, providing the much-needed tissue context [39]. For a detailed overview of collection methods, we refer the readers to several recent comprehensive reviews [38,40,41,42,43].
RNA extraction is not recommended for single-cell material because of the limited RNA concentration [44]. Cells are rather collected directly into lysis buffers, where they burst in the hypotonic environment and release their intracellular content. At this stage, RNases are not a principal threat to the released RNA. The more numerous extracellular RNases are washed away during preparation of the cell suspension and during cell collection. Intracellular RNases are, on the other hand, quickly diluted in the lysis buffer volume and/or inactivated by the lysis buffer components and low working temperatures [45]. Similar to RNA extraction, DNase treatment is not recommended as it increases the sample volume and dilutes the RNA. Amplification of the genomic DNA (gDNA) is prevented in the design of qPCR primers, which target exons separated by intron(s) of a substantial length (Section 5.1 Assay Design) [27,44].
Cells are collected directly into the lysis buffer. Besides rupturing the cellular membrane, the lysis buffer also acts as a stabilizing agent, protecting RNA from degradation and its adherence to plastic walls. Although the literature and companies offer multiple complex candidates, a simple solution of 0.1% BSA in nuclease-free water (NFW) maintains a high RNA quality even for extended storage time periods at room temperature (up to four hours) or repeated freezing and thawing [45]. An alternative approach is to collect cells directly into a reverse transcription buffer. Reverse transcription buffers should be, however, used with caution in terms of lysis and storage efficiency, as the performance will depend on the particular kit used.
Suitable capture and storage units for collected single cells are 96- or 384-well RNase- and DNase-free plastic plates. Not only can they be readily used in the next steps, but the processing of the entire plate also provides statistically adequate cell numbers for data analysis. The collected single-cell material should be stored at −80 °C. The plates need to be sealed properly with hardback foils manufactured to withstand low temperatures. Although occasional freezing and thawing of single cells should not hamper the sample quality for scRT-qPCR analysis, we raise caution against this practice [46,47].
Single cells do not necessarily need to be sampled individually per collection vessel but may also be collected in small bulks numbering tens to hundreds of cells. Although small bulk analysis does not provide a true single-cell resolution, it increases the RNA concentration and allows the detection of transcripts that are present in just a few copies per cell (e.g., lncRNAs, transcription factors) [28,48,49]. The limiting factor of small bulk analysis is the maximum number of cells per lysis buffer volume ensuring proper cell lysis and dilutions of intracellular RNases as well as other potential inhibitors [45]. The maximal recommended number of cells collected per collection vessel depends on the cell type, volume, and properties of the lysis buffer as well as the volume of the co-transferred medium. A typical cell-to-volume ratio ensuring good results is ~100 cells per microliter of lysis buffer [45].
Reverse transcription (RT) is an enzymatic reaction where RNAs are reverse transcribed into complementary DNA (cDNA) sequences ( Figure 3 ). RNA molecules that fail to be transcribed are omitted from downstream processing steps and are not detected. This makes RT the primary bottleneck of the entire RT-qPCR workflow.
RT is often referenced as the least reliable step of the scRT-qPCR workflow [20,22,50,51,52,53,54]. This is attributable to multiple options for how RT can be performed. RT performance is dependent on the selection of: (i) the reverse transcriptase (RTase) [51,52,55,56,57]; (ii) primers [27,53,55,58,59]; (iii) additives [55,60,61]; and (iv) temperature profile [56,62,63,64,65,66]. Additionally, minute amounts of starting material (tens of picograms of RNA per single cell) add a layer of complexity related to the stochasticity of the quantification of small copy numbers [28,51,54,55,67].
Key RTase parameters are absolute sensitivity and reproducibility, which are closely tied to the reaction efficiency. These and other parameters have been recently examined in a broad spectrum of RTases by two independent studies [56,61]. These studies showed that Maxima H- and SuperScript IV (both ThermoFisher) are currently the most efficient RTases, due to which they are recommended for single-cell applications. Moreover, both RTases are applicable not only in scRT-qPCR but also in scRNA-Seq thanks to their template switching activity. Their performance is supported by their high processivity, increased thermostability, high synthesis rate, robustness to inhibitors, and RNase H and strand displacement activity. High processivity enables reverse transcription of long transcripts (declared up to 20 kb). Increased thermal conditions loosen secondary structures, a known impairment to cDNA synthesis [63,65,68,69]. The higher synthesis rate shortens the experimental time and increases the RTases’ turnover. Insensitivity to inhibitors allows for the introduction of additives, broadening the RTases’ applicability. Strand displacement activity in combination with the lack of RNase H enables the synthesis of multiple cDNA molecules using a single RNA molecule as a template [56]. Template switching activity is a central property in several scRNA-Seq protocols, allowing the incorporation of additional sequences to the cDNA ends, which serve as barcodes and/or further cDNA amplification [68,70].
Although RTases represent a key reaction component, RT cannot be efficiently initiated without a primer. Priming also has an essential role in the reaction efficiency and specificity. In practice, two priming methods are used in scRT-qPCR: (i) 3′end-oriented oligo(dT), or (ii) random hexamers having the potential to prime the reaction from any site on the RNA. Oligo(dT) priming is selective for polyadenylated transcripts. However, this selectivity comes with the risk of 3′end bias, as the RTase may stall due to secondary structures or bound proteins preventing transcription of the 5′ end [53,65,68]. This should be, therefore, considered in the qPCR primer design. Random hexamers lack the selectivity for RNA types, are less sensitive to RNA quality (they do not rely on the presence of a polyA tail which is first affected by degradation), and can prime single RNA molecules from multiple sites [53,56,57]. To achieve a balanced transcript coverage, the two priming methods are often combined for superior performance. Combinations of both priming strategies with high-end performing RTases can deliver RT yields of over 100% [55,56,71]. Finally, increased primer concentrations were shown to increase reaction yields in bulk samples [58,59]; however, this effect was not studied for single-cell material.
Various compounds have been shown to increase the RT efficiency, e.g., tRNA [53], total extracted RNA [58,60], MgCl2, and betaine or trehalose [72]. Surprisingly, recent reports failed to reproduce their enhancing properties, questioning the further application of these additives [55,61]. Alternatively, physical separation of the reaction into smaller reaction chambers has been shown to increase the reaction efficiency [73]. Locally increased reagent concentrations and faster assembly of molecular machineries accelerate the rate of interactions, which results in an improved RT efficiency. As this approach is hardly applicable on the already small volumes of single-cell measurements, the molecular crowding effect may be delivered by the addition of polyethylene glycol (PEG 8000), a mechanism already implemented in recent scRNA-Seq protocols [61,74].
The majority of RT protocols consist of two steps. Firstly, mixtures of dNTPs, spike-ins, and RT primers are added to the sample and incubated at an elevated temperature. The high temperature melts the secondary structures and ensures that the primers can anneal more equally along the template length. Therefore, it is not recommended to omit this step as it may negatively affect the reaction efficiency [56]. In the second step, an RT buffer, RNase inhibitor, and RTase are added. The use of RNase inhibitors is recommended to minimize the risk of lower reaction yields due to the activity of the contaminating RNases.
Good laboratory practices represent a key aspect to deliver reproducible and reliable results. Before RT mastermixes are prepared, all reagents need to be vortexed and spun down, except for protein-based reagents (RNase inhibitors and RTases). Protein-based reagents are only briefly spun down, and, except for during the pipetting time, they are best kept at the recommended storage conditions. dNTPs and primers are typically prepared in aliquots that minimize repeated freeze–thawing and are supplied in concentrations ready for use. Once prepared, the RT mastermix is added directly to the sample, and by doing so, one avoids the unnecessary transfer of single-cell material. Minimal RT volumes are preferred as they maintain high concentrations of the target molecules and reagents, improving the outcome of scRT-qPCR [44,51,61,68,73,75]. The use of RNase-free plastic is a necessity [65]. Biological replicates are preferred over technical replicates [22,23]. cDNA is stored at 4 °C for up to 24 h, but for long-term storage, −20 °C is preferable.
RNA spike-ins represent an effective tool for quality control throughout the entire RT-qPCR workflow [76]. RNA spike-ins are molecules of an artificial and unique sequence that are added in equal amount into each RT reaction and consequently measured by a dedicated qPCR assay. A uniform signal across all the samples shows that none of the samples were subjected to pipetting error or inhibition. Commercial spike-in kits differ in the complexity of their contents. Simple options contain a single RNA transcript variant (e.g., TATAA RNA Spike-in, TATAA Biocenter), whereas more complex alternatives are composed of a panel of RNA transcripts varying in length and abundance (e.g., ERCC Spike-in, ThermoFisher), mimicking the transcriptome complexity [77]. The latter option is more suitable for applications such as scRNA-Seq, where the total number of screened assays is not a limiting factor. Being a quality control reagent, it is important to prevent a spike-in’s repeated freeze–thawing; thus, storing it aliquoted, similar to dNTPs and primers, is advised. As a rule of thumb, the measured Cq signal for a spike-in assay should not deviate by more than one cycle from the mean Cq spike-in signal per plate.
Single-cell RT is performed in low volumes to prevent unnecessary RNA dilution and loss of sensitivity [22]. This implies that only a limited sample volume is available for follow-up qPCR measurements. Quantification of multiple targets per cell therefore requires their enrichment by preamplification ( Figure 4 ). Noteworthy, preamplification (preAMP) is not mandatory for targeted profiling of a limited number of transcripts where sample dilution is not applied.
Targeted preAMP is a regular multiplex PCR, but instead of a single pair of forward and reverse primers (further referred to as assay), a multiplicity (tens) of assays are employed in the reaction at once. Typically, assays used in targeted preAMP are identical to those used in the later qPCR step. The pool of assays is selected to address the experimental question of interest, enriching the reaction contents exclusively for cDNAs of targeted transcripts [78,79]. cDNA transcripts that are not amplified by the primer assay pool cannot be quantified in subsequent qPCR as they are diluted to a non-detectable concentration in the further steps. An alternative to target-specific preAMP is global preAMP, where a single pair of primers is used to amplify cDNA copies globally for all transcripts [72,80]. The flexibility of global preAMP is, however, paid off in a reduced yield and sensitivity and increased variability [80]. Due to these reasons, global preAMP is not frequently used in scRT-qPCR experiments.
Well-optimized preAMP is a result of several fine-tuned parameters including: (i) individual assay efficiency; (ii) number of assays per preAMP assay pool; (iii) concentration of the preAMP assay pool in the reaction; (iv) reaction volume; (v) temperature profile; and (vi) the number of amplification cycles. Well-optimized preAMP aims to deliver a reproducible output for each assay [23]. An assay employed in scRT-qPCR needs to be target-specific, produce no primer dimers, not amplify gDNA, and showcase a reproducible efficiency of >90% (see Section 5 Quantitative PCR for further details) [22,27,81]. Although efficient in separate assays, the pool of preAMP assays has potential to give rise to unspecific products. Still, more complex preAMP assay pools (>50 assays) produce fewer artefacts than those combining less assays, deliver higher yields, and are less variable [79]. This is attributed to the increased number of interactions that make the formation of stable primer dimers less likely. The formation of unspecific products is also dependent on the primer concentration and number of cycles. To minimize their production, the recommended number of cycles is between 18 and 22 cycles, and the final primer concentration should be around 40 nM, which is about ten times less than in standard single-plex PCR [79]. Lastly, to account for annealing heterogeneity of the preAMP assay pool, the annealing temperature is set as the average of the respective assays (optimally within ± 1 °C of the assays’ annealing optima), and the annealing time is prolonged to three minutes [79].
Before any preAMP assay pool is employed, it is necessary to inspect whether its individual assay components maintain a high level of performance even when combined in the assay pool. Screening of more assays provides a more encompassing overview, but screening of the entire assay list would also prove to be very laborious and an expensive task. In our practice, a subset of assays targeting cell type markers, genes of interest, and spike-ins (minimum of eight in total) is sufficient for general assessment of the preAMP performance. Validation is performed on random cDNA, gDNA, and non-template control (NTC) samples by inspecting the cycle of quantification (Cq) for non-preamplified and preamplified samples [81]. The first control is focused on the variability of deltaCq (SD∆Cq), where deltaCq is the difference in Cq between non-preamplified and preamplified cDNA. The decision on the degree of tolerated variability is left to the researcher; however, its value should approach the standard qPCR variability (SDCq) of ~ 0.2 among replicates [22]. In our perspective, a preAMP assay pool with assays scoring SD∆Cq ≤ 0.5 is considered robust and reliable. A step-by-step guide on how to calculate SD∆Cq can be found in the Supplementary Material. Secondly, the measurement of gDNA provides insight into the sensitivity of each assay to the gDNA background before and after preAMP. Thirdly, the NTC sample reports on the production of unspecific products and serves for the identification of assays that may report false positive signals in preamplified samples.
After validation of the preAMP reaction, the finalized reaction setup begins with the selection of assays of interest. Apart from the assays determined in the experimental design, the assay list ought to include control assays as well, e.g., a spike-in assay or an assay to control the gDNA content [82]. The preAMP assay pool is prepared on ice and stored in aliquots to avoid primer degradation during freeze–thaw cycles. Any PCR mastermix can be used, although those containing more enzymes are preferred. The selected PCR mastermix does not need to include fluorescence reporters, as the reaction is not visualized by a standard. Undiluted cDNA samples are typically mixed with a mastermix in a 1-to-9 ratio. Here, 10× cDNA dilution is mandatory to prevent PCR inhibition by RT reagents [54,58,83,84]. Although it is theoretically possible to process an entire cDNA sample and thus achieve maximal sensitivity, for cost reasons, only a portion of single-cell cDNA is routinely analyzed. This consequently determines the portion of the single-cell transcriptome analyzed in the RT-qPCR workflow. After the preAMP reaction, the preamplified material needs to be immediately diluted and frozen at −20 °C to terminate polymerase activity and production of potential artefacts.
Quantitative PCR (qPCR) is the third and final laboratory step of the scRT-qPCR workflow ( Figure 5 ). It extends the previous preAMP step, as the assays are usually shared for the preAMP and qPCR steps. Contrary to preAMP, the qPCR performance is visualized after every cycle and is typically run in simplex (single assay per reaction) using non-specific dyes such as SybrGreen or EvaGreen.
