NEW YORK – A new report by researchers from the Chinese University of Hong Kong has brought to light comprehensive information about the mechanisms that define how intact DNA from cells is sheared into the patterns of small fragments that circulate in the blood.
According to authors, the study, appearing in this week's issue of the American Journal of Human Genetics, has resolved what were previously unanswered questions about this process, which could be useful in the optimization of current liquid biopsy methods and development of new avenues for blood-based medical testing.
"We know that cfDNA is in short fragments but not much has been known about how those fragments are being created," lead author Dennis Lo said in an interview. "So basically, what this paper does is analyzes step by step the various nucleases that are responsible for creating [DNA] fragments."
In the study, Lo and colleagues explored the roles of three enzymes — DNASE1, DNASE1L3, and DNA fragmentation factor subunit beta (DFFB) — using mouse models deficient in each of these nucleases, hoping to piece together a model that could describe the fragmentation process that generates cfDNA molecules.
By sequencing the mouse blood samples and comparing the fragment ends of cfDNA molecules in each type of nuclease-deficient mouse to those in wild-type mice, the group was able to demonstrate that each nuclease in question has a specific cutting preference that suggests a stepwise process of cfDNA fragmentation.
Based on what they saw, the researchers concluded that cfDNA fragments appear to be generated intracellularly at first, in the presence of DFFB, intracellular DNASE1L3, and other nucleases, and then extracellularly under the influence of circulating DNASE1L3 and DNASE1.
According to Lo and his coauthors, the work definitively links the action of distinct nucleases to the cfDNA fragment end profiles observed in cfDNA, "clarifying the fundamental biology and biography" of how these fragments are produced.
Having a precise picture of this fragmentation process could open up new applications for cell-free DNA testing, Lo added in an interview. For one, knowing how the process of DNA fragmentation is supposed to take place, investigators may be able to find new links between aberrant fragmentation and particular pathological consequences or disease states.
In addition, as cfDNA fragmentation has started to become a feature of various methods being developed for cancer early detection, information about the biology of this process could aid in validation and benchmarking of these methods.
Speaking at a recent scientific meeting, Johns Hopkins investigator Viktor Velculescu mentioned this need directly, describing methods he and his colleagues have developed that analyze cfDNA fragmentation patters to detect early cancer.
He noted at the time that a challenge to comprehensive technical validation of such methods is a lack of established models that describe what nucleases are involved in digesting the genome to create cfDNA.
Lo said that the new work also offers important new information about how preanalytical variables such as anticoagulant type and time delay in blood separation may confound cfDNA analysis.
The experiments the group performed to create their DNA fragmentation model relied on a method of incubating whole blood in EDTA to induce apoptotic cell death in vitro in a way that might mimic the process of cfDNA fragmentation and dissolution in an actual organism.
Blood from a normal mouse treated in this way and then sequenced revealed longer cfDNA molecules enriched for A-end fragments, in particular, A< >A, A<>G, and A< >C fragments with "a strong nucleosomal periodicity at 200 bp and 400 bp," the authors wrote.
In contrast, using blood from DFFB-deficient mice there was no such enrichment. According to the authors, this implies that DFFB is likely responsible for generating these A-end fragments.
Shoring up the hypothesis, Lo and colleagues cited prior literature supporting DFFB as playing a major role in DNA fragmentation during apoptosis.
"Enzyme characterization studies have shown that DFFB creates blunt double-strand breaks in open internucleosomal DNA regions, and that this process has a preference for A and G nucleotides. This biology of blunt double-stranded cutting only at internucleosomal linker regions would explain the nucleosomal patterning in A< >A, A< >G, and A< >C fragments," the researchers wrote.
After establishing the impact of DFFB, the team's work was not complete though, because the investigators could see that the typical profile of actual observed cfDNA is still very different from the profile produced by their EDTA-incubation process.
Typical cfDNA obtained before incubation trends toward C ends across all fragment sizes, as opposed to the A-end enrichment observed. Thus, there had to be one or more other nucleases involved, likely doing their work after the generation of newly released A-end fragments.
Because they had seen that C-end fragment predominance is lost in DNASe1L3-deficient mice, the group turned to this enzyme as the likely culprit, but they wanted to investigate whether its activity takes place intracellularly or in circulation after initial fragments are released.
Conducting additional experiments, they saw evidence of what they describe as a two-step process, with activity of DNASE1L3 appearing to occur in both settings.
Finally, the team also sought to understand how a third enzyme, DNASE1 may also be acting on these processes.
In previous work, the group found that cfDNA fragment size profiles didn't seem to differ between DNASE1-deficient and wild-type mice. But according to the authors, DNASE1 is known to require proteolytic help for its activity.
The group decided to use the blood thinner heparin to try to mimic the function of in vivo proteases. With heparin incubation, they were able to see an increase in T-end fragments in samples from wild-type versus deficient mice, "predominantly among subnucleosomally sized (50– 150 bp) fragments, suggesting that DNASE1 has a role in generating short <150 bp fragments."
"We believe that this first model has included a number of key nucleases involved in cfDNA generation, but the model can be further refined in the future. For example, other potential apoptotic nucleases include endonuclease G, AIF, topoisomerase II, and cyclophilins, and there are probably more to be discovered," the team wrote.
The authors cautioned that their in vitro incubation system likely fails to capture the full complexity of the in vivo reality. However, they argued that "the insights generated should nonetheless be valuable."
One important takeaway, Lo said, is that in having a clearer model of how this process occurs, researchers now also have important information about things that could disturb it, including how long samples are stored in tubes, and whether they are exposed to heparin — something that, based on the team's study, would be expected to dramatically alter the cfDNA fragment profile of a sample.
Admittedly, not knowing about these potential confounders hasn't prevented people from developing predictive signatures that use cfDNA fragmentation patterns, whether for things like non-invasive prenatal testing, or in cancer.
In a way, Lo said, this new schema helps illustrate how lucky the field has been to be successful so far, "without knowing all these mechanisms."
"Now, knowing it, that will just help us do all of this much better," he argued.
Finally, the team's model of a progressive, stepwise DNA fragmentation process could enable new research seeking to link changes in this normal process to other diseases or disorders.
"It's possible that certain conditions could alter nuclease levels, and that could be a very exciting thing," Lo said. In such a case, the group's fragment end analysis method could be applied to human samples, with the prevalence of certain nucleotides in various end positions showing which specific nuclease is or isn't playing its proper role.