The manuscript describes a chip-based digital PCR assay to detect a rare CDH1 transcript variant (CDH1a) in fresh-frozen normal and tumor tissues obtained from patients with gastric cancer.
CDH1a, a non-canonical transcript of the CDH1 gene, has been found to be expressed in some gastric cancer (GC) cell lines, whereas it is absent in normal gastric mucosa. Recently, we detected CDH1a transcript variant in fresh-frozen tumor tissues obtained from patients with GC. The expression of this variant in tissue samples was investigated by the chip-based digital PCR (dPCR) approach presented here. dPCR offers the potential for an accurate, robust, and highly sensitive measurement of nucleic acids and is increasingly utilized for many applications in different fields. dPCR is capable of detecting rare targets; in addition, dPCR offers the possibility for absolute and precise quantification of nucleic acids without the need for calibrators and standard curves. In fact, the reaction partitioning enriches the target from the background, which improves amplification efficiency and tolerance to inhibitors. Such characteristics make dPCR an optimal tool for the detection of the CDH1a rare transcript.
The CDH1 gene encodes for E-cadherin, a key factor involved in the maintenance of the normal gastric epithelium through the regulation of cell adhesion, survival, proliferation, and migration1. Loss of E-cadherin protein as a result of deleterious germline or somatic alterations of CDH1 has been associated with the development of GC2,3. Non-canonical transcripts arising from intron 2 of the gene have also been hypothesized to play a role in gastric carcinogenesis4,5. In particular, one such transcript, CDH1a, has been shown to be expressed in GC cell lines but is absent from the normal stomach4. We recently detected CDH1a in GC tissue samples from GC patients using chip-based dPCR5. dPCR was used to evaluate, for the first time, the presence of the CDH1a gene transcript in intestinal GC and in normal tissue.
The gold standard method to determine gene expression is Real-Time quantitative PCR (qPCR). However, the resulting data can sometimes be variable and of poor quality, especially when the level of target in the sample is low. This variability can be caused by contaminants, which inhibit polymerase activity and primer annealing, leading to non-specific amplification and competitive side reactions6.
Although the basic biochemical principles of dPCR are similar to those of qPCR, dPCR shows some advantages, allowing for very precise measurements of genomic DNA (gDNA)/complementary DNA (cDNA) molecules. Indeed, dPCR is an end-point reaction that relies on the calibrated partitioning of a sample into thousands of wells, so that each well contains zero or a single target molecule. Amplification then occurs only in the wells containing a copy of the target and is indicated by a fluorescent signal. The absolute number of target molecules in the original sample can then be calculated by determining the ratio of positive to total partitions using binomial Poisson statistics7.
In addition, the dPCR technique eliminates the need for running a standard curve and, hence, the associated bias and variability, allowing for a direct quantification of targets8,9; it produces more precise and reproducible results independently of contaminants and efficiency due to its high tolerance to inhibitors10; it is more sensitive and specific than qPCR, and is thus a reliable method for the detection of a rare target. Finally, the partitioning of the sample into multiple reactions reduces the competition with background molecules and improves the limit of target detection, making amplification possible and facilitating the detection of single molecules of gDNA/cDNA6. The detection and quantification of nucleic acids by chip-based dPCR has been increasingly applied to copy number variation, quantification of DNA fragments, and mutation analyses11,12,13, given the precision and low material input requirement of the method. In addition, dPCR has recently been integrated into the analysis of both microRNAs14 and gene transcripts5,15.
The protocol follows the guidelines of the IRST Human Research Ethics Committee.
NOTE: This procedure is specifically designed for the detection of a low number of cDNA molecules in human fresh-frozen tissues. The tissue sections have been cut on dry ice, while still frozen, from previously validated patient-derived gastric tumor or normal tissue samples.
1. RNA Isolation and Purification
2. cDNA Synthesis
3. Digital PCR Reaction Set Up
4. Data Analysis and Interpretation
Using the procedure presented here, we checked for the expression of the rare transcript variant CDH1a in gastric fresh-frozen tissues. The analysis by dPCR was performed on 21-paired normal and cancer tissue samples and in 11 additional tumor samples. CDH1a was detectable in 15 out of 32 (47%) tumors, whereas no normal tissue samples showed the presence of this rare transcript5. In our analysis, chips with less than 13,000 data points were rejected, as were chips with non-homogeneous loading. Figure 1 shows a chip view for both evaluable (Figure 1A) and non-evaluable (Figure 1B, C) samples. In the case of the latter, the chip must be discarded due to the presence of multiple bubbles (Figure 1B), a single big bubble (Figure 1C), or as a result of an insufficient number of data points above threshold (Figure 1B). These samples must be run again to obtain interpretable results. With regard to the scatter plot provided by the software (Figure 2), it normally depicts the signals from the FAM reporter dye (Y-axis) against signals from the VIC (manufacturer's proprietary) reporter dye (X-axis). The data points presented on the scatter plot are color-coded, and here, we only visualized the FAM reporter dye signals in blue (Figure 2A), indicating the wells in which the target was amplified. These signals are located closer to the Y-axis and further from the origin of the plot. The second color seen on the plot is yellow and is indicative of the wells in which no amplification occurred. A reliable threshold for the positive signals was set and applied to all the chips to prevent call bias. Here, the selected threshold was 6,000 based on the range of signals observed for the various chips in these experiments. CDH1a cDNA copies (blue signals) were found in a number of tumor samples, showing a clear difference in amplitude from negative signals (Figure 2A). Conversely, no signals above the selected threshold were found in samples negative for CDH1a transcript, i.e., all normal tissue samples and some tumor samples (Figure 2B). Similarly, no positive signals were detected in the NTC in which water was added instead of cDNA, hence serving as a negative control for the dPCR reaction (Figure 2C).
A fine chip analysis must be applied to filter the low-quality data points and eliminate the risk of ambiguous results. This was done by viewing the chip to determine the exact location of the corresponding positive signal: if the blue signal is located at the very borders of the chip (Figure 3A) or around bubbles (Figure 1B), the signal is dismissed from the analysis. To further prevent the risk of false positives, signals above 6,000 for the FAM channel and on the far right of the plot were considered as probably aspecific and were thus filtered out, even though they were localized well within the chip (Figure 3B). In this way, a sample is considered positive for the expression of the rare transcript, CDH1a, even if it shows a single amplification signal, as long as that signal complies with the aforementioned conditions.
Figure 1. Representative evaluable and non-evaluable chips. The blue dots are wells in which the CDH1a transcript was amplified; the yellow dots are wells in which no amplification occurred; the white dots are empty wells. Data points above threshold are indicated underneath each chip. (A) An evaluable chip with blue signals localized well within the chip's borders. (B) A non-evaluable chip with less than 13,000 data points above threshold. (C) A non-evaluable chip with a large bubble. Please click here to view a larger version of this figure.
Figure 2. Representative digital PCR scatter plots of CDH1a expression. Plots depict signals from the FAM reporter dye (Y-axis) against signals from the VIC reporter dye (X-axis). The blue dots are CDH1a positive expression signals, while the yellow dots are the "No amplification" signals. (A) CDH1a-positive sample. (B) CDH1a-negative sample. (C) No template control (NTC) showing zero CDH1a expression. Please click here to view a larger version of this figure.
Figure 3. Representative fine-chip analysis of digital PCR scatter plots. On the left is the chip view with the signals of interest zoomed-in in a black square, while on the right is the corresponding scatter plot. (A) A blue signal localized on the correct fraction of the plot but at the border of the chip. (B) Three blue signals highlighted in grey localized on the far right of the plot but within the chip. Please click here to view a larger version of this figure.
dPCR was originally developed for DNA molecular measurements10,11,12,13 and in time this technology was adapted for the quantification of microRNAs and RNA transcripts5,14,15. In this protocol we have extended the list of applications to include detection of rare transcripts derived from fresh-frozen tissue samples. To that purpose we utilized a rather high amount of cDNA (300 ng). Although this amount is not the maximum that can be used in dPCR, it was sufficient for setting up standardized conditions to compare all our samples. By setting the florescence signal threshold at 6,000 and minding the locations of positive amplification signals on both the chip and scatter plot, we successfully detected this rare target in gastric tumor tissue samples.
By dPCR, we could not absolutely quantify CDH1a, due to a very low amount of the target, but we could reliably assess the presence/absence of this rare transcript. Indeed, the quantitative nature of this technique is limited by the initial number of targets present in the analyzed samples7. Of relevance, the technique itself remains to be fairly expensive and time intensive. Thus, when the analysis of numerous samples and/or of several targets is required, alternative techniques should be considered7,8.
That being said, dPCR undoubtedly provides the ultimate platform for sensitive measurement and quantification of nucleic acids, as it improves precision and reproducibility with respect to qPCR7,8. Moreover, it could be considered as the only method available for analyzing gene expression of low-abundant nucleic acids that near the limit of qPCR sensitivity6.
The protocol here described can be adapted to RNAs from any tissues, enabling the detection of rare targets, as in our study. In addition, the assay could be extended onto absolute quantification of transcripts. In such cases, the number of copies/µL reported by the instrument for a given sample must be simply multiplied by the loaded reaction volume and divided by the initial amount of cDNA. To overcome variability produced by reverse-transcription, gene-specific calibrators should be included, when possible6. Looking to the future, dPCR holds considerable potential in becoming the gold standard platform, not only for rare sequence detection, but also rare mutation detection and precise copy number quantification7. This is especially relevant in genetic mosaicism, as well as in cancer research where tumor heterogeneity can impact a patient's clinical outcome and response to treatment.
From the practical point of view, when setting up the dPCR protocol there are multiple critical steps that must be highlighted. First of all, the reaction mix containing the appropriate amount of cDNA based on the target expression level expected, should be transferred with great care in the loading blade so as not to create air bubbles, which could interfere with sample homogeneity. In addition, the blade itself should first be positioned appropriately; otherwise it could result in an uneven sample distribution. Furthermore, sufficient coating with immersion fluid and accurate sealing of the lid should always be performed. Regarding chip processing and analysis, it is important to be aware of the possibility of accumulating condensation on the chip; in such a scenario, the chip should be re-wiped with isopropanol and re-analyzed. Moreover, as dPCR is based on Poisson statistics, a minimum number of evaluable data points (>10,000) must be present on the chip, otherwise the sample should be run again. Finally, for low target copy detection experiments, a critical confounding factor could be the signal-to-noise ratio, but a reliable threshold positioning can help in identifying real positive signals.
The authors have nothing to disclose.
The authors wish to thank Gráinne Tierney for editorial assistance.
TRIazol Reagent | Thermo Fisher Scientific | 15596018 | |
Glycogen 20 mg/ml | ROCHE | 10901393001 | |
RNeasy MinElute Cleanup kit | QIAGEN | 74204 | |
iScript cDNA Synthesis kit | BioRad | 1708891 | |
QuantStudio 3D Digital PCR Master Mix v2 | Thermo Fisher Scientific | A26358 | |
CDH1a IDT custom designed assay | Integrated DNA Technologies (IDT) | NA | F) GCTGCAGTTTCACTTTTAGTG (R) ACTTTGAATCGGGTGTCGAG (P)/FAM/CGGTCGACAAAGGACAGCCTATT/TAMRA/ [dPCR optimized assay concentrations: 900 nM (F), 900 nM (R), 250 nM (P)] |
QuantStudio 3D Digital PCR 20K Chip Kit v2 | Thermo Fisher Scientific | A26316 | |
Heraeus Biofuge Fresco | Thermo Scientific | 75002402 | |
Thermocycler (Labcycler) | Sensoquest | 011-103 | |
GeneAmp PCR System 9700 | Thermo Fisher Scientific | N805-0200 | |
Dual Flat Block Sample Module | Thermo Fisher Scientific | 4425757 | |
QuantStudio 3D Tilt Base for Dual Flat Block GeneAmp PCR System 9700 | Thermo Fisher Scientific | 4486414 | |
QuantStudio 3D Digital PCR Chip Adapter Kit for Flat Block Thermal Cycler | Thermo Fisher Scientific | 4485513 | |
QuantStudio 3D Digital PCR Chip Loader | Thermo Fisher Scientific | 4482592 | |
QuantStudio 3D Digital PCR Instrument with power cord | Thermo Fisher Scientific | 4489084 |