NEW YORK (GenomeWeb) – Researchers from the Karolinska Institute in Stockholm have developed an amplification-free method for sequencing cell-free DNA, and demonstrated that it is as accurate as amplification-based methods for noninvasive prenatal diagnosis of aneuploidy.
The researchers also showed that the method provides greater and more even coverage and less bias in GC rich regions, which would be important for analyzing genomic regions that are typically challenging to sequence.
The team recently published details of the method in the journal Genomics.
Sten Linnarsson, senior author of the paper and an associate professor at the Karolinska Institute, told GenomeWeb that he is interested in using the method clinically both for prenatal diagnostics and also in an oncology setting to assess cell-free DNA of cancer patients.
The method was adapted from a protocol developed by Max Planck researchers to sequence ancient DNA, Linnarsson said. Although that method did use amplification, many features of ancient DNA sequencing also apply to sequencing cell-free DNA. For instance, in both cases, DNA is often fragmented, single-stranded, and is of limited quantity, Linnarsson said.
The method starts with a denaturation step so that all the DNA fragments are single-stranded. Then, the team binds the fragments to biotinylated linkers and attaches them to streptavidin beads. Binding the DNA fragments to beads helps prevent DNA loss in the subsequent steps.
Next, researchers add the first adaptor, which is single-stranded and consists of a section that is complementary to the linker bound to the streptavidin beads and a section that can be used in the subsequent sequencing. A polymerase extends the adaptor, creating a double-stranded complex. Then researchers ligate a second adaptor to the complex. After that, the strands are removed from the beads and ready for sequencing.
One key difference in the amplification-free protocol compared to the ancient DNA protocol is that the adaptors are added by ligation, as opposed to PCR, Linnarsson said. In addition, he said the group had to make a few other optimizations to prevent DNA loss. For instance, the second adaptor includes a 4-bp overhang that prevents it from ligating directly to the first adaptor. Otherwise, the adaptor-adaptor complex would make up a large portion of the sequencing library.
The team compared the amplification-free method to a standard PCR-based method, and also to a protocol that uses unique molecular identifiers (UMI), which is useful for counting applications. Linnarsson's group previously developed the UMI method for single-cell transcriptome sequencing to directly count transcripts.
The team tested the method on 31 clinical samples of maternal plasma. Fifteen donors had a fetus with least one chromosomal aberration and 16 were healthy. Of the 16 healthy samples, four were used for normalization.
From 50 ng of starting sample, the team was able to generate 400 million reads.
For comparison, they performed the PCR-based method on two additional healthy samples and the UMI method on a trisomy 21 male sample.
The amplification-free method accurately called all trisomies, however, in one case it called a sample as XY, contradicting quantitative fluorescence-PCR results that called it as XXY. To solve the discrepancy, the researchers obtained genomic DNA from the fetal lung and tested it with array CGH and the amplification-free method. Both cases confirmed that the sample was mosaic, where about 30 percent of cells were XY and 70 percent were XXY.
To evaluate library prep for each method, the team looked at the variance in read density between bins both before and after normalization. Prior to normalization, the PCR-free and UMI methods had less variance than the PCR-based method, which was expected. Somewhat surprisingly though, the methods had comparable variance after normalization.
"When the protocols are used for calling chromosomal copy number, the biases introduced by PCR are there but are highly reproducible," Linnarsson said. "For that kind of large-scale quantification, both methods are equally good."
However, the PCR-free approach would be preferable "in situations where it's not easy to find a good control," he said.
Because certain areas of the genome are difficult to amplify — for example in areas of high GC content — the PCR-free approach gives better, more even coverage of the genome, Linnarsson said.
The method will likely have the most application in situations "where it's important to cover everything," such as when "looking for specific mutations or small microdeletions," Linnarsson. "You don't want to have holes there."
When the team compared performance of the methods in areas of low GC content, they found that the PCR-based method "recovered almost no molecules with a GC content less than 12 percent," while the amplification-free method recovered low GC content molecules "reasonably well." On average, the authors wrote, in regions where GC content was 5 percent below the average, the amplification-free method showed 5 to 10 percent higher coverage.
Linnarsson said that while he does not currently have plans to commercialize the method, he is open to the idea. Although he has not done a cost comparison, he said that the PCR and PCR-free methods should be comparable. "The major cost [for both methods] is the DNA sequencing and the cost of obtaining the sample," he said. "In principal there might be some savings [in the PCR-free method] in that you could sequence a little less, but we didn't demonstrate that."
Turnaround time and labor are also similar, he said, adding that the PCR-free method would be amenable to automation since it is done on magnetic beads.
In the immediate future, he said he is interested in collaborating with other groups to apply the method in a clinical setting for noninvasive prenatal screening.
His group is also looking at applying the method to sequencing cell-free DNA from cancer patients. Using an amplification-free method to sequence circulating cell-free tumor DNA may have even more benefits than for NIPT, because it "is very important to have very accurate coverage of the whole genome," Linnarsson said. "In cancer genomes, you're looking at amplifications and deletions of pieces of chromosomes, so it's important to have an accurate view."