Multiplex Digital PCR Assay Targeting DNA Methylation for Early Detection of Cancer

Abstract

Digital PCR has been demonstrated to enable the precise nucleic acid absolute quantification across a wide range of applications. Historically, dPCR technology was limited to two channels for fluorescence detection. Recent advances in digital PCR have made it possible to multiplex and allow for simultaneous analysis of multiple targets within a PCR reaction. Here, we describe a multiplex dPCR assay for the detection and quantification of methylated DNA (also known as DNA-methylation-specific multiplex digital PCR, or DNA-methylation specific dPCR). We also demonstrate the application of this technology for the development of a candidate DNA methylation-based biomarker panel for the early detection of high-grade dysplasia and esophageal adenocarcinoma. Further, we discuss common technical challenges and trouble-shooting for performing successful DNA-methylation-specific MS-dPCR assays.

1. Introduction

DNA methylation is an epigenetic modification that occurs when a methyl group is covalently added to a cytosine base of DNA, typically at cytosine-guanine (CpG) dinucleotides. DNA methylation is one of the most common epigenetic modifications and plays an important role in regulating gene expression. Unsurprisingly, abnormal changes in DNA methylation have been implicated in pathological processes or diseases, such as cancer. Aberrant DNA methylation patterns in cancer tissue, when compared to normal counterparts, can contribute to the entire tumorigenesis process and have been recognized as one of the hallmarks of cancer [1]. Consequently, DNA methylation-based biomarkers hold the promise for early detection of cancer.

Digital PCR (dPCR) technology is an advancement in quantitative PCR (qPCR) for the quantification of nucleic acids. In contrast to traditional qPCRs, where PCR is performed in a bulk reaction, dPCR involves collecting data from a large number of individual reaction chambers, termed partitions. Recent advances in nanofluidic techniques and emulsion chemistries have allowed the generation of tens of thousands of partitions of uniform nanoliter size, thus greatly enhancing the capacity of dPCR technology for the detection and absolute quantification of rare molecular events with unprecedented precision. The quantification depends on an independent distribution of target molecules into equally sized partitions, meaning that the number of targets in each partition follows a standard Poisson distribution [2]. This partitioning system gives dPCR superior sensitivity over other iterations of PCR. It has been demonstrated that the analysis method used in dPCR is less affected by reaction inhibitors because it uses endpoint measurements, whereas qPCR relies on amplification kinetics [3]. Therefore, current dPCR technologies have the potential to have an impact on biomarker applications [4, 5]. While early detection biomarker development has benefited from the increase in sensitivity and precision offered by the dPCR approaches, most digital assays for DNA methylation have been limited to single- or duplex-assays for the detection of one or two targets in one reaction. When it comes to a biomarker panel that is often composed of multiple targets, the assessment of each biomarker by single- or duplex assays in clinical samples can be labor-, time-, and costintensive, in addition to the fact that many types of precious clinical samples are difficult to obtain and are only available in small quantities. Therefore, there is an unmet need to develop multiplex PCR assays for DNA methylation-based biomarker development.

Here, we describe the application of dPCR technology for the detection and quantification of methylated DNA (also known as DNA-methylation-specific multiplex digital PCR, or MS-dPCR) using the QIAcuity dPCR system (Qiagen), which achieves partitioning through nanoplates. We present a quadraplex dPCR assay for a candidate DNA methylation-based biomarker panel for the early detection of high-grade dysplasia and esophageal adenocarcinoma. Furthermore, we discuss special considerations in the development of a biomarker panel for the early detection of cancer using MS-dPCR assays, in addition to common technical challenges and troubleshooting needed to perform successful experiments and fully unleash the potential of this technology. See Fig. 1 for a diagram of the workflow.

Fig. 1.

Workflow. Workflow diagram for the Methylation-specific digital PCR protocol

2. Materials

2.1. Assay Components and Controls

1. Bisulfite converted Methylated and Unmethylated control DNA, from the EpiTect PCR Control DNA Set (catalog #59695) (see Note 1).

2. Primer Express^® V3.0.1 (Life Technologies, catalog #4363991). This software can be used to predict the secondary structure of the assay primers and probes and to visualize their Tm and %GC content.

2.2. Sample Preparation Extraction of gDNA

1. For DNA isolated from fresh frozen tissue, we recommend the use of the DNeasy^® Blood and Tissue Kit (QIAGEN, catalog #69504). For FFPE tissue, we recommend the QIAamp^® DNA FFPE Tissue Kit (QIAGEN, catalog # 56404).

2. DNAse and RNase free microcentrifuge tubes, 1.5 mL.

3. Vortex and microcentrifuge.

4. Centrifuge.

5. Molecular grade water.

6. 100% ethanol.

7. Xylene (for FFPE paraffin removal).

8. A heat block or water bath.

2.3. Bisulfite Conversion

1. Molecular biology grade water.

2. EZ DNA methylation Kit (Zymo Research, Catalog # D5002).

3. Strip tubes.

4. A thermocycler, incubator, or other device that allows for a constant temperature overnight.

5. Microcentrifuge tubes 1.5 mL (DNase and RNAse free).

6. Microcentrifuge.

7. Centrifuge.

2.4. DNA Quantification

1. QuantiT^™ PicoGreen dsDNA Assay Kit (Life Technologies, p7589).

2. 96-well Black/Clear Falcon^™ 96-Well, Cell Culture-Treated, Flat-Bottom Microplate (Fisher Scientific, cat#353219).

3. DNAse and RNase-free microcentrifuge tubes, 1.5 mL.

4. A microcentrifuge and vortex.

5. Molecular grade water.

2.5. Qiagen QIAcuity Consumables and Equipment

1. 26 k 24-well, QIAcuity Nanoplate (catalog # 25003).

2. Nanoplate Seals (catalog #250099).

3. QIAcuity Probe PCR Kit containing QIAcuity Probe Master-mix and water (catalog # 250102) (see Note 2).

4. Forward and reverse primers.

5. Probes to label each target.

6. QIAcuity One 5plex Device (catalog # 911021).

7. QIAcuity Nanoplate Tray (catalog #250098).

8. QIAcuity Software Suite.

9. Nanoplate roller.

10. Electronic multichannel pipette.

3. Methods

3.1. Extraction of gDNA from Clinical Samples

1. Genomic DNA (gDNA) isolation can be performed using either the Qiagen’s DNeasy^® blood and tissue kit, or the QIAamp^® DNA FFPE kit. Select the appropriate kit depending on the starting material and follow the manufacturer’s instructions.

2. DNA can be quantified using the Quant-iT^™ PicoGreen^® dsDNA Assay Kit according to the manufacture instructions (see Note 3).

3.2. Bisulfite Conversion of gDNA

Bisulfite conversion can be performed using a kit such as the EZ DNA methylation kit (Zymo Research) following the manufacturer’s instructions with the modifications included in the protocol below. Note that after bisulfite conversion, the gDNA is transformed in single-stranded fragments, which will be partitioned in dPCR.

1. Prepare the CT conversion reagent by combining 750 μLof water and 210 μL of the kits M-Dilution Buffer to the tube of CT Conversion Reagent. Mix by vortexing for 10 min.

2. Add 5 μL of M-Dilution Buffer to the DNA sample and use water to adjust the total volume to 50 μL.

3. Incubate the sample for 15 min at 37 °C in a thermocycler or heat block.

4. Once the incubation is over, add the 100 μL of previously prepared CT Conversion Reagent to the sample and mix by pipetting.

5. Incubate for 12–16 h at 50 °C in the dark then hold at 4 °C until ready to proceed, using a water bath, thermocycler, or heat block. Sample can remain at 4 °C for up to 20 h.

6. Next add 400ul of M-Binding Buffer the Zymo-Spin^™ IC Column and place it into a collection tube.

7. Load your sample from step 5 into the Zymo-Spin^™ IC Column with buffer prepared in step 6.

8. Centrifuge at >10,000 g for 30 s, then discard the flow through.

9. Wash the column using 100 μL of M-Wash Buffer. Centrifuge at >10,000 gfor 30 s.

10. Add 200 μL of M-Desulfonation Buffer to the column and incubate at room temperature for 15–20 min. After the incubation centrifuge at >10,000 g for 30 s.

11. Add 200 μL of M-Wash Buffer to the column. Centrifuge at >10,000 g for 30 s. Add 200 μL of M-Wash Buffer and centrifuge at >10,000 g for 30 s.

12. After the final wash, spin-dry the column for an additional 1 min (total of 2 min).

13. Place the column into a 1.5 mL microcentrifuge tube. Add 10 μL of M-Elution Buffer directly to the column. Centrifuge for 30 s at >10,000 g to elute the DNA. To optimize the DNA yield, perform two elutions of 10 μL into the same collection tube.

14. Bisulfite converted DNA can be used immediately or stored at −80 °C for long-term storage. Single-use aliquots are recommended to avoid repeated freeze-thawing of the converted DNA.

3.3. Methylation-Specific dPCR Assay Design Guidelines

1. The maximum length of the amplicon is 160 bp for optimal performance, with a preferred size of 80–100 bp.

2. The bisulfite-converted sequence of either the top or the bottom strand can be used as the input sequence in the PyroMark Assay Design Tool.

3. As a rule, each primer and probe sequence should contain more than two CG oligodeoxynucleotides (CpGs). For primer design, CpGs should be located near 3′ end of the primers but try to avoid repeating more than 3 C or G mononucleotide at the end of the sequence. For probe design, the CG of interest should be located in the middle or near 3′ end of the probe. If only one CpG can be found (e.g., in a CpG sparse region), the CpG should be positioned near the 3′ end in the primers or in the middle of the probe to improve the specificity toward methylated alleles. Additional specificity can be achieved by using minor groove binding (MGB) probes. Multiple target analysis is possible using multiplexing (see Note 4).

3.4. QIAcuity Digital PCR Assay Setup and PCR Cycling

1. All procedures are to take place at room temperature unless otherwise noted. Pre-and post-PCR procedures need to occur in designated rooms, such that the pre-PCR steps can occur in a clean room free of post-PCR products (see Note 5).

2. Assemble the PCR reaction using the correct proportion as described in Table 1, which contains enough of the PCR reaction mix for two technical replicates.

3. Use an electronic multichannel pipette to transfer 38 μL of the reaction mixture from the strip tubes to the QIAcuity plate. Dispense the volume onto the wall of the well in the QIAcuity plate, being careful not to introduce bubbles into the well. Bubbles, as well as defects in the plate or damage to the plate, can cause dead space in the partition data.

4. Once the QIAcuity plate has been loaded, it needs to be sealed properly. Follow manufacturer instructions to seal the plate and then to transfer it to the QIAcuity machine for cycling and reading following conditions described in Table 2. Be careful not to tilt, bump, or drop the plate during transfer.

Table 1.

Master mix composition and calculations in dPCR reactions

PCR reaction mix component	Starting	Final	Volume (μL) for 1 well
QIAcuity Probe Mastermix	4×	1×	10
Methylated target primer probe mix	10×	1×	4
C-LESS-C1 primer probe mix	10×	1×	4
RNase-free water	–	–	20
Total volume	–	–	38

Table 2.

Thermocycling conditions for dPCR

Step	Time	Temperature	Number of cycles
Heat activation	2 min	95 °C	1×
Denaturation	15 s	95 °C	40×
Annealing/extension	1 min	60 °Ca, b

^aAnnealing temperature varies for each assay. Use a temperature gradient to establish optimal annealing temperature

^bThe annealing temperature for C-LESS-C1 is 60 °C

3.5. Data Analysis and Interpretation

1. After the run and image transfer have been completed, the QIAcuity Software Suite can be used to analyze the data (see Note 6). First, check for any error messages, such as one that notes some wells are oversaturated. If this message appears, follow the manufacturer’s guide to adjust the gain and exposure appropriately and reimage the plate.

2. Check partition images in the signal map tab for the reference channel. Note any areas of dead space. Any large regions of dead space in the data may impact the results and should be documented (Fig. 2).

3. Look at the partition data in the signal map for the target channels; the image should show a random distribution of positive partitions (Fig. 3).

4. A threshold is used to determine positive and negative partitions. Rely on non-template control (NTC) and negative control samples to set the proper threshold level (Fig. 4).

5. After setting the threshold for each assay, the data can be exported into Excel software for further analysis.

6. First, calculate the RM% for each individual sample replicate separately using the formula below.

   $$\text{RM}\%=\left(\frac{\text{Methylation Assay}\left(\frac{\text{copies}}{\mu \text{L}}\right)}{C-\text{LESS}-C1\left(\frac{\text{copies}}{\mu \text{L}}\right)}\right)\times 100$$

7. Average the RM% of the replicates to get the average RM% of the sample.

8. Calculate the RM% standard deviation between replicates (Fig. 5).

9. Next, use the average (μ) and the standard deviation (σ) of the RM% to calculate the coefficient of variation (%CV) for the replicates.

   $$\%\text{CV}=\frac{\sigma }{\mu }\times 100$$

10. Finally, screen the data for any samples with a %CV of greater than 30; consider rerunning these samples if DNA is available.

Fig. 2.

QIAcuity Reference channel partition data. Representative images of the QIAcuity reference channel from a MS-dPCR assay. Each Row (A and B) represents a data from and individual QIAcuity nanowell. The blue in each row represents the reference dye that flows into the partitions when the plate is rolled. Row A shows the ideal result; each square is blue indicating the reference dye was distributed into every partition evenly. Row B is a representative depiction of incomplete filling of the reference dye leading to areas of dead space, which are indicated by the red arrow and circle. Dead space partitions will not be analyzable, and a large amount of dead space will impact the quality and reproducibility of the final data

Fig. 3.

QIAcuity fluorescence channel partition data. A representative image of a QIAcuity fluorescence channel, with the random distribution of positive partitions shown in green

Fig. 4.

QIAcuity 1D plot. 1D plot of C-LESS-C1 in the EpiTect controls run in duplicate. The controls appear in the following order; 100% EpiTect methylated (M) (wells C2 and D2), 100% Epitect unmethylated (UM) (wells E3 and F3), and a NTC (wells G3 and H3). The 1D plot shows the separation of positive partitions (navy) and negative partitions (gray). The red line is the threshold that has been set, between the positive and negative partitions. The threshold setting is done based on the non-template control (NTC). The threshold is set so that the baseline noise of negative partitions in the NTC well is negative

Fig. 5.

MS-dPCR LOQ 1D-plot. 1D plot for a MS-dPCR assay showing representative results of a limit of quantification (LOQ) experiment, in which the 100% methylated Epitect control was diluted down into a background of the 100% unmethylated Epitect control for a final total DNA input of 8 ng. The positive signal (blue) is determined by the threshold (red line), which is set above the negative partition baseline (gray) and the baseline noise from the non-template control (NTC) well (G3). For each fourfold dilution of the 100% Epitect methylated control, a corresponding decrease in positives signal detected as positive partitions for the MS-dPCR assay is observed

3.6. Consideration on MS-dPCR Panel for Early Cancer Detection

3.6.1. Integrity and Amount of Input DNA

Working with archived clinical specimens poses unique challenges for the development of DNA-methylation-based biomarker assays. First, the integrity of input DNA from FFPE tissue blocks is suboptimal due to the chemical modification from formalin cross-linking, denaturation, and fragmentation during the bisulfite conversion process, in addition to the potential PCR inhibitors present in clinical samples. Therefore, even though a dPCR assay may be both precise and reproducible when measuring methylated genes in reference samples, its performance may become suboptimal in DNA extracted from FFPE blocks. Second, due to the precious nature of the clinical samples, the availability of input DNA can become the major limiting factor, thus requiring a highly sensitive assay. In this regard, dPCR offers advantages over qPCR to overcome these barriers and achieve the precision needed to develop a robust biomarker assay for clinical samples.

3.6.2. Choice of Fluorescence Channels on the dPCR Platform

Designing an assay for the detection and absolute quantification of multiple rare, methylated alleles from tumor DNA in a background of unmethylated DNA from non-tumor DNA is technically constraining. A robust and specific assay for each target is the most crucial of the criteria that must first be met before a multiplex panel of assays can be considered. Once the assays for individual targets have been worked up, there are further technical considerations that need to be considered. Probe synthesis itself can be another design constraint when the user considers assigning a fluorophore to an assay. Manufacturers have specifications that may limit what length of probe can be synthesized for a specific fluorophore, and MGB’s which give added specificity to an assay are not available for all fluorophores. One strategy is to place the most robust assay in the darkest channel (Cy5), and the least robust one in the brightest channel. Of specific consideration to the QIAcuity’s platform, among the five available channels, it is likely that only four (green, yellow, red, and crimson) can be successfully multiplexed as the fifth channel (orange) is known to leak out into other channels.

3.6.3. Plate Design

1. To ensure a successful experiment using methylation-specific dPCR assays, it is important to carefully perform plate design to include proper controls for data analysis and interpretation. Both positive and negative controls are crucial for all PCR assays, including dPCR. Three types of control samples must be used along with the experimental samples.

   1. A positive control for methylation detection assays. This positive control can be commercially available, e.g., 100% methylated EpiTect control DNA (Qiagen), or from reference tumor DNA that is known to be methylated in the CpGs of interest. The use of the positive control is needed for initial assay development and performance evaluation (LOD and LOQ). A positive control with a defined amount can also serve as a calibrator to include on each plate, a valuable tool to assess run-to-run variability and to perform data analysis.
   2. A negative control for methylation detection assays ensures that the assay is methylation-specific. This negative control can be commercially available, e.g., 100% unmethylated (UM) EpiTect control DNA (Qiagen), or from reference samples that are known to be unmethylated in the CpGs of interest.
   3. No-template control (NTC) detects any contamination in the reaction mix that might lead to false positive reactions. The NTC can be used to set the threshold to distinguish positive from negative partitions.
   4. Inclusion of a control reaction (C-LESS-C1) in a multiplex assay to simultaneously measure the amount of amplifiable DNA in a sample, in addition to the target of interest (Table 3), is highly recommended.

Table 3.

Primers and probe sequences for the C-LESS-C1 assay

C-LESS-C1	Primer/probe sequence (5′–3′)	Reference
Forward primer	TTGTATGTATGTGAGTGTGGGAGAGAGA	[6]
Reverse primer	TTTCTTCCACCCCTTCTCTTCC
Probe	VIC-CCTCCCCCTCTAACTCTAT-MGBNFQa

^aMGBNFQ refers to a Minor Groove Binder (MGB) nonfluorescent quencher at the 3′ end of the probe. This quencher is essential for the use of a 60 °C annealing temperature with C-LESS-C1 (see Note 7)

3.6.4. Reaction Optimization

Since dPCR relies on data collection from a large number of independent partitions, in which a single reaction occurs, reaction optimization is particularly important for a successful dPCR run. In dPCR reactions using the QIAcuity instrument, an error message from wells reported with oversaturated signals is usually an indicator that too much probe is being used for the assay. On the other hand, if the concentration of either the primers or the probe is too low, the dynamic range of a dPCR assay is suppressed. Therefore, it’s crucial to conduct a preliminary assessment for primers/probe concentration, after a careful primers/probe design, to achieve the maximum signal from a PCR reaction without saturating the signal with too much probe. For multiplex assays, the whole reaction optimization process should be performed, and the performance of a multiplex assay should be compared to single-plex reactions. For trouble shooting issues with assay performance and optimization please see Table 4.

Table 4.

Troubleshooting

Observation	Explanation	Suggestion
No signal	Incomplete reaction mix assembly	Check the input volumes of each component and the concentration added to the master mix
Large areas of dead space	Improper plate handling	Take care not to drop or bump the plate during transfer and loading
Bisulfite converted DNA has a loss of signal (copies/μL) over time	Bisulfite converted DNA degradation due to freeze-thaw cycles between uses	Aliquot the bisulfite-converted DNA for single, double, or triple use, and discard after three freeze-thaw cycles
Loss of probe amplitude over time	Probe degradation due to light exposure and/or freeze-thaw cycles	Aliquot the stock probe into single, double, or triple use aliquots
The negative baseline is below 30 relative fluorescence units (rfu)	The PCR reaction is being limited by probe availability	Increase the probe concentration
Data comes back and is reported to be oversaturated	Excess probe is being used in the reaction	Test a gradient of probe concentrations to optimize probe input

3.6.5. Evaluation of Assay Performance

To ensure the performance of MS-dPCR, it’s important to understand and appropriately determine both the limit of detection (LOD) and the limit of quantification (LOQ). While the LOD is defined as the lowest copies of methylated DNA that can be accurately and reliably quantified by the assay using water as a diluent, the LOQ is a more comprehensive parameter to consider since it indicates an assay’s capacity to detect abnormal methylation from tumor cells in a background of unmethylated DNA from normal cells, with acceptable accuracy and precision. While an assay’s LOD can be used to guide determination of LOQ, extrinsic factors such as the availability of input DNA from clinical samples can influence an LOQ of a biomarker assay; i.e., LOQ can be improved with increasing input DNA. Both LOD and LOQ should be determined using the dilution series with 100% methylated, and 100% unmethylated EpiTect controls as background diluents for LOQ (Fig. 5). For all multiplex assays, we utilize the signal from C-LESS-C1 to account for the total amount of DNA in the sample regardless of methylation status. This allows us to use the control DNA to create ratios of mixtures of methylated and unmethylated DNA and report the RM% for each mixture, and to see at what point the expected ratio of methylation diverges from the calculated RM%.

3.6.6. Experimental Replicates

To increase the chance of detecting rare methylated DNA from tumor cells, a common strategy is to pool multiple wells (hyper-well) and generate larger numbers of partitions than single wells. For clinical samples with limited availability, we suggest a minimum of duplicate wells, but triplicates or quadruplicates could be used to achieve desirable LOQ.