The concept of digital PCR (dPCR) for nucleic acid detection and quantification is not a new one. In 1992, Sykes et al. described a method in which they measured the number of initial DNA targets in a sample rather than PCR-amplified products (Sykes et al. 1992). Using limiting dilution, PCR, and the Poisson distribution, this helped them quantitate the absolute number of leukemic cells in a leukemia sample. A few years later, in 1999, the term dPCR was used to describe a method introduced by Vogelstein and Kinzler in which samples were partitioned and amplified so that the resulting PCR products were either completely mutant or completely wild type (Vogelstein and Kinzler 1999). Acknowledging the limitations of PCR in the detection and verification of low-level mutations, the authors tested the feasibility of their approach using stool samples from patients with colorectal cancer to detect a mutant KRAS oncogene. They found that dPCR could be used to detect mutations present at relatively low levels.
Digital PCR offers a highly sensitive alternative to conventional qPCR, enabling absolute quantification of nucleic acids and rare allele detection. The method is based on the partitioning of a DNA or cDNA sample into thousands of discrete subunits, prior to PCR amplification. DNA molecules are distributed such that only a portion of the subunits contain molecules of target DNA. Each subunit is then treated as an individual PCR reaction and the DNA within is amplified by end-point PCR using target-specific primers and fluorescent probes or dye. Following amplification, the subunits are analyzed for the presence (positive) or absence (negative) of fluorescence signal. The ratio of positive to negative subunits is calculated and the Poisson distribution is used to determine the absolute quantity of initial target DNA or RNA in the sample.
The original methods from the 1990s have been modified and a number of different dPCR techniques now exist. These include using droplet partitions based on oil-water emulsions (Hindson et al. 2011), PCR amplification on a microfluidic chip (Fan and Quake 2007, Ottensen et al. 2006, Warren et al. 2006), and separation onto microarrays (Morrison et al. 2006) or spinning microfluidic discs (Sundberg et al. 2010). Though dPCR is the preferred method for detecting and quantifying nucleic acids in many cases, it is a technique where experience counts. To that end, the following tips will help you get started and optimize your dPCR experiments when using droplet partitioning technology.
Know How Much Input DNA to Use
The amount of input DNA will vary depending on the digital PCR system you use. The range for Bio-Rad’s Droplet Digital™ PCR (ddPCR™) System is 1 to 100,000 total copies of target DNA per well. This amounts to between 3.3 pg and 350 ng of human genomic DNA (gDNA). The sweet spot is 30,000 copies per well, where the variance is the lowest (Ottensen et al. 2006). For other organisms, genome size per copy can be calculated. For very small genomes, such as bacteria and viruses, several dilutions may be necessary to get within the 1–100,000 total copy range. For gene expression experiments, the amount of input cDNA will depend on the expression level of the targets. Serial dilutions are also helpful, not only to get into the correct range, but also to see the linearity of your input DNA.
Concentrate FFPE Samples before Amplification
When working with highly degraded DNA, such as formalin-fixed paraffin-embedded (FFPE) samples, there may be a large difference between the amount of quantifiable DNA (measured using a Nanodrop Spectrophotometer or Qubit Fluorometer) and the amount of intact, amplifiable DNA. Based on empirical data, only approximately 40% of FFPE DNA is amplifiable. Due to this, loading a greater amount of DNA is a viable option. However, since FFPE samples usually contain inhibitors, sometimes loading less volume of DNA can also help. Hence, to truly maximize experimental success, we recommend concentrating the DNA sample beforehand.
Lower Ramp Rate for Cleaner Data
The droplets in the PCR reaction are immobile, decreasing rates of normal aqueous thermal diffusion. Thus, lowering the ramp rate to 2 degrees per second on the thermal cycler ensures a more uniform thermal transfer to all of the droplets. For certain assays, this can result in cleaner data with improved separation due to the even thermal cycling of all droplets.
Modify PCR Conditions for Better Amplification
The PCR cycling conditions can be changed if working with difficult templates.
- For longer amplicons (>400 bp), change the protocol from a two- to three-step by adding a 72°C extension cycle for 1–6 minutes, depending on the length of the amplicon
- Most bacteria and viruses can go straight into droplets without the need for sample preparation and DNA isolation. Change the PCR protocol from 10 min at 95°C to 10 min at 98°C to lyse the bacteria or virus within the droplets
- For GC-rich templates, try changing the PCR cycling conditions from 94°C for 10 sec to 96°C for 10 sec, for the usual 40 cycles, to help with amplification
Find One True Positive Sample in a Million
To determine whether a sample is a true positive, we recommend the Rule of 3. Determine the false positive rate (FPR) from the no template controls (NTCs). Next, multiply the number of positive droplets per well by three. The positive samples should have at least three times the number of positive droplets than the FPR.
To detect a rare, for instance one in a million, species of DNA, 7–10 μg of DNA will need to be screened depending on the type of sample. Here’s how to figure out how much DNA to screen. To find one rare target in 1,000 background molecules, or 0.1% sensitivity, use the Rule of 3. You would want to see at least three positive droplets in 3,000 background molecules for statistical significance. 3,000 copies of a haploid genome is equivalent to 1,500 cells or approximately 10 ng of DNA (Morton 1991). Hence, for 0.0001% sensitivity, or 1 in a million, screen approximately 10 μg of DNA.
Understand What Error Bars Mean
In ddPCR Systems, the error bars for each well represent the 95% confidence intervals using Poisson statistics and the total number of droplets. If merging wells, for technical triplicates for instance, two error bars will appear. The outer one represents the standard error of the mean (SEM) for the replicates while the inner error bar is the 95% confidence interval for the Poisson distribution, which is now based on the total number of droplets from the three wells.
Digest Samples for More Accurate Quantification
Digestion using a restriction enzyme is useful when using more than 66 ng of gDNA due to its structural complexity. When working with plasmids, using restriction digestion to linearize the plasmid allows more access to the supercoiled DNA and accurate quantification of the plasmid. In addition, digestion is important for copy number analysis to ensure that tandem repeats are separated. Digestion is also recommended when working with FFPE samples to ensure that all degraded FFPE samples are digested to the same extent.
Digestion can be carried out directly in the ddPCR Supermix. Use high-fidelity restriction enzymes and make sure they do not cut within the amplicon. If working with a particularly difficult template, such as a GC-rich template, try a four- vs. six-cutter restriction enzyme.
Use Multiple Reference Genes
When assessing copy number variation (CNV), particularly in cancer samples, it is important to ensure that the reference assay is stable at a count of 2 copies per genome. Often, the reference assays themselves are amplified or deleted in cancer samples. We recommend trying multiple reference genes and offer four standard assays and 57 pericentromeric assays to ensure that the CNV calls are correct. Screening eight reference assays with your target of interest will ensure that four will be concordant 99% of the time and reflect the true CNV of the sample (unpublished data).
As the scientific community continues to demand greater data accuracy and more credible results, technologies such as dPCR, which can detect and quantify low-level nucleic acids, are becoming more and more commonplace. Following the advice above will help you address some of the common issues associated with setting up and running experiments and analyzing the resulting data when using droplet partitioning technology in dPCR.
Fan HC and Quake SR (2007). Detection of aneuploidy with digital polymerase chain reaction. Anal Chem 79, 7,576–7,579.
Hindson BJ et al. (2011). High-throughput droplet digital PCR system for absolute quantitation of DNA copy number. Anal Chem 83, 8,604–8,610.
Morrison T et al.(2006). Nanoliter high-throughput quantitative PCR. Nucleic Acids Res 34, e123.
Morton NE (1991). Parameters of the human genome. Proc Natl Acad Sci USA 88, 7,474–7,476.
Ottesen EA et al. (2006). Microfluidic digital PCR enables multigene analysis of individual environmental bacteria. Science 314, 1,464–1,467.
Sundberg SO et al. (2010). Spinning disk platform for microfluidic digital polymerase chain reaction. Anal Chem 82, 1,546–1,550.
Sykes PJ et al. (1992). Quantitation of targets for PCR by use of limiting dilution. Biotechniques 13, 444–449.
Vogelstein B and Kinzler KW (1999). Digital PCR. Proc Natl Acad Sci USA 96, 9,236–9,241.
Warren L et al. (2006). Transcription factor profiling in individual hematopoietic progenitors by digital RT-PCR. Proc Natl Acad Sci USA 103, 17,807–17,812.
First posted on biocompare.com on November 7, 2016.