NEW YORK (GenomeWeb) – Researchers at Sequenom have demonstrated that whole-genome bisulfite sequencing can help distinguish between placental and maternal circulating cell-free DNA.
Reporting in Genome Biology last week, the team performed whole-genome bisulfite sequencing on eight pregnant women and seven non-pregnant women, as well as on seven genomic samples isolated from maternal buffy coat and five placenta samples.
The work could have implications not only for noninvasive prenatal testing, but also for oncology for detecting epigenetic changes in circulating tumor DNA.
The goal of the study was to "establish the baseline of methylation in circulating cell-free DNA, before exploring and potentially utilizing epigenetic differences in medically relevant applications," Dirk van den Boom, Sequenom's chief scientific and strategy officer, told GenomeWeb.
Although van den Boom said that the study would likely not have any impact on the company's current noninvasive prenatal test offerings, including its MaterniT21 Plus test and its VisibiliT test, he said that the study could potentially have relevance for future tests, although he did not disclose details.
"Cell-free DNA that is circulating in plasma can originate from essentially any tissue in contact with blood," he explained, which is important in the context of both pregnancy and cancer, where the placenta and tumor tissue contribute significant genomic material, respectively, van den Boom added.
Previous groups have explored whole-genome bisulfite sequencing of cell-free DNA, including researchers from Dennis Lo's lab at the Chinese University of Hong Kong who developed a bisulfite sequencing approach to detect methylation in circulating fetal DNA. Separately, a group from the Garvan Institute of Medical Research in Darlinghurst, Sydney, published on a method to sequence methylated circulating cell-free DNA using a modified ChIP-seq approach and are working on identifying a methylation signature in the ccfDNA that can identify women with ovarian cancer.
Van den Boom said that although Lo's study showed methylation differences between the fetus and the mother, Sequenom wanted to "establish a full catalog" of those differences.
The group first performed whole-genome bisulfite sequencing on ccfDNA from the eight non-pregnant women, generating about 10x coverage of 74 to 92 percent of the 28 million or so genomic CpG sites. Approximately 75 percent of all CpG cytosines were methylated, while methylation elsewhere in the genome was minimal. They used these eight methylome maps to form the foundation for comparisons.
Because previous work has found that the main source of ccfDNA in non-pregnant women are hematopoietic in origin, the team performed whole-genome bisulfite sequencing on buffy coat samples — which contain white blood cells — from seven pregnant women to see if the same methylation patterns held true. They found a similar distribution pattern of methylation in those samples as from non-pregnant women. By examining data from ENCODE, the team was able to confirm that the majority of ccfDNA in non-pregnant women is of hematopoietic origin.
However, they did find 152 differentially methylated regions, indicating that there are other contributors to ccfDNA in non-pregnant women aside from hematopoietic cells. While the exact contribution is not known, the authors speculated that it could come from organ systems that have significant contact with the bloodstream, including the kidneys, liver, or endothelium.
Next, they performed whole-genome bisulfite sequencing on five placenta samples to identify placenta-specific differentially methylated regions. In contrast to the non-pregnant women, which had minimal intermediate methylation at CpG sites, occurring at only 15.5 percent of CpG sites, in the placental samples, nearly 47 percent of CpG sites displayed intermediate methylation. The vast majority of the CpG sites that were intermediately methylated in placenta were highly methylated in both the non-pregnant ccfDNA and buffy coat samples.
In addition, they identified 51,259 differentially methylated regions between pregnant and placenta samples and 105,874 differentially methylated regions between buffy coat and placenta.
Looking at the genomic distribution of the differentially methylated regions "uncovered large contiguous genomic regions with significant placental hypomethylation relevant to non-pregnant ccfDNA," the authors wrote.
Those regions of hypomethylation also had a number of distinguishing characteristics, including being in regions of low CpG and with few genes. In addition, the regions were often highly methylated in non-pregnant women. Although more work is needed to characterize those large swaths of placental hypomethylation, the characteristics that the researchers identified may "underscore a lack of heterochromatin formation during early placenta development or allele-specific methylation of regions with low CpG density in the placenta," the authors wrote.
Finally, the team performed whole-genome bisulfite sequencing of ccfDNA from the seven pregnant women. Compared to non-pregnant women, there was significantly less methylation at CpG sites. By examining the mean methylation levels at CpG sites in the differentially methylated regions between non-pregnant ccfDNA and placental DNA, the group confirmed that the difference in methylation levels between non-pregnant and pregnant women was due to the contribution from the fetus.
Interestingly, the group also found that the size of ccfDNA corresponds to its methylation status. Previous reports have found that fetal ccfDNA is shorter than its maternal counterpart. One reason for that, the researchers theorized, could be because hypomethylation tends to lead to an open chromatin structure, which could provide better access for endonucleases to cleave the DNA into shorter fragments.
Evaluating DNA fragment length and CpG sites, the team found that CpG cytosines within fragments longer than 200 bp were on average 12.3 times more likely to be methylated.
Finally, the researchers demonstrated that they could use methylation status to enrich for fetal DNA from a ccfDNA sample. They analyzed samples from 12 pregnant donors, three of which carried a fetus positive for trisomy 21, both without enriching for unmethylated DNA and enriching for unmethylated DNA. Enriching for the unmethylated DNA resulted in a nearly four-fold increase in chromosome 21 z-scores in the trisomy 21 samples. However, one euploid sample also had similar enrichment as the trisomy 21 samples, so if enriching for unmethylated DNA had been used to determine trisomy, it would have been classified as a false positive.
"While these data are promising as an early proof of concept, further work is needed to evaluate the robust performance of DNA hypomethylation as a method for fetal DNA enrichment in ccfDNA derived from the plasma of pregnant donors," the authors wrote.
Van den Boom explained that incorporating such a strategy into a noninvasive prenatal test would be relatively straightforward. For instance, "methylation-sensitive restriction enzymes can be used to digest non-methylated DNA and leave the methylated DNA intact," he said. Alternatively, "methods could employ proteins that specifically bind to methylated DNA," which could then be used to "fish out" the methylated DNA, leaving behind the hypomethylated fetal DNA.
Despite the potential of incorporating such a method into an NIPT, van den Boom said that Sequenom would likely not do so for its existing tests. However, he said, analyzing ccfDNA also holds potential in oncology.
"Epigenetic changes are not only common but also biologically important in the development of cancer," he said. The study lays the "foundation for further exploration of epigenetic differences in circulating cell-free DNA and their potential use for diagnostic purposes in various diseases."
Although the method may have applications in oncology, van den Boom declined to disclose whether the company would incorporate methylation differences into any of its planned tests in that space, such as a liquid biopsy test that it plans to launch in the second half of the year.