BALTIMORE – Researchers at Cornell University and their collaborators have developed a new metagenomic DNA sequencing approach for cell-free DNA (cfDNA) samples that promises to be robust against environmental DNA contamination introduced during sample preparation.
The method, named Sample-Intrinsic microbial DNA Found by Tagging and sequencing (SIFT-seq) and described in a Nature Communications paper published last month, leverages bisulfite salt treatment to chemically tag cfDNA in the sample prior to DNA extraction, enabling researchers to discern and eliminate downstream foreign DNA contamination using bioinformatic processing.
“The method was developed to tackle an important problem in metagenomics, that is, environmental DNA contamination,” said Iwijn De Vlaminck, a biomedical engineering professor at Cornell University and the lead investigator of this study. According to him, environmental DNA contamination, which can be introduced through reagents, the laboratory environment, human handling, and other means, can especially be pernicious for studies where the sample microbial biomass is very low.
“You might even have scenarios where there is no microbial DNA present in the actual sample where all the signals coming from contamination,” De Vlaminck added. “This is an issue for many microbiome studies.”
To help tackle the challenge, De Vlaminck’s team came up with a scheme to chemically label the sample cfDNA upstream of the sample preparation process prior to DNA extraction. “The idea is to tag the DNA in the sample as early as possible,” he explained, adding that the labeling step allows researchers to flag untagged foreign DNA that are introduced during sample preparation.
While there are several potential strategies to label the DNA, De Vlaminck’s group settled on using bisulfite conversion, where unmethylated cytosines within the DNA are deaminated. De Vlaminck noted there are a few advantages of using bisulfite salt treatment for the labeling step.
For one, he said the method can work efficiently in biofluid samples such as blood and urine, allowing researchers to tag the DNA as upstream of the workflow as possible.
In addition, he said bisulfite conversion, which is purely a chemical reaction, does not require enzymes or oligos that often already come with trace amounts of environmental DNA contaminant due to the manufacturing process.
For this study, De Vlaminck’s team benchmarked the performance of SIFT-seq with six proof-of-principle applications — spanning from identifying viral and bacterial COVID-19 co-infection to detecting urinary tract infection — in blood and urine samples. In general, the results showed that when applied to cfDNA in plasma and urine, SIFT-seq can reduce background signals from common contaminants by up to three orders of magnitude.
“The data is quite strong,” De Vlaminck said, noting that the results demonstrated SIFT-seq as “quite a powerful way” to deal with DNA contaminations.
“One of the really nice things about the method in this paper is that the intervention is pretty far upstream in the process; that allows you to look at contamination almost from the very start,” said Daryl Gohl, a researcher at the University of Minnesota Genomics Center who is experienced with next-generation sequencing method development.
Gohl said compared with many previously developed contamination mitigation approaches that focus on the library prep stage, SIFT-seq has the advantage that it enables researchers to intercept environmental DNA contamination even during DNA extraction.
Despite the method’s promises, Gohl said one of the concerns he has with bisulfite conversion is how it will interfere with the quantitative accuracy when it comes to detecting the levels of different microbial species in the sample. “Bisulfite conversion can lead to damage and degradation of the DNA can also lead to amplification biases,” he explained. “Clearly, [the method] is helping with the binary — presence and absence — of the contamination. But if you want to apply this more in a research setting to characterize the composition of a low-biomass microbiome, then those quantitative questions will become more important.”
Still, Gohl said even if bisulfite treatment ends up being problematic for certain applications, researchers can explore using other DNA labeling schemes while deploying the same framework described in this paper. “I think this is a really important paper for introducing that concept to the field,” he added.
“This paper is really incredible,” said Laura Weyrich, a professor at Pennsylvania State University whose lab predominantly works with low-biomass microbiome samples that are prone to environmental contamination. “It provides a new method for us to be able to look at very low-scale pathogen detection.”
While Weyrich considers SIFT-seq “absolutely exciting,” one of the downsides of the method, she said, is that it only deals with cfDNA, which exists outside of the microorganism. “If you contaminate a sample with some other living microbes, then you wouldn’t be able to use this method because the DNA is still protected in the cell walls,” she pointed out.
Additionally, Weyrich said it remains to be seen whether the method will work with all samples, especially those with incredibly low prevalence of DNA, given that bisulfite treatment tends to be “very harsh” on the DNA and might require high DNA quantities to go through the process.
Mirroring Weyrich’s point, Wei Wang, director of the sequencing and microarray facility at Princeton University, said he hopes to see more data on the effect of bisulfite conversion on the sample DNA. “I feel what’s lacked [in this paper] is how much [DNA] loss will incur at the expense of bisulfite treatments, because bisulfite [can cause] quite severe damage and degradation to DNA samples,” he said.
In addition, because bisulfite treatment can reduce the DNA sequence diversity by converting four bases to three bases, Wang said more data are needed to evaluate the impact of SIFT-seq for distinguishing closely related bacteria species within the sample.
To that end, De Vlaminck acknowledged that “a major drawback” of labeling DNA with bisulfite treatment is that it reduces the DNA alphabet from four letters to three letters. Although he said the altered genetic code can still provide enough information to identify particular microbial species, it can lower the resolution to distinguish certain genetic features, such as single nucleotide variants.
Another limitation of SIFT-seq, De Vlaminck noted, is that because the method is only robust against DNA contamination introduced after the labeling step, it cannot detect contamination that occurred during the sample collection or isolation of the plasma from whole blood.
De Vlaminck said that DNA damage is less of an issue for DNA that already has a low molecular weight, adding that the team has observed “limited additional degradation due to bisulfite conversion, because the DNA is already very short.”
Although this study did not apply SIFT-seq in solid sample types, De Vlaminck said in theory the method should work on these samples, as well. He said the team has also submitted a patent application related to the method.
As for the cost, he said because bisulfite chemicals are “not overly expensive” and SIFT-seq’s sequencing coverage is very similar to conventional sequencing, the method should add “minimal costs” to standard NGS experiments.