SAN FRANCISCO (GenomeWeb) – Researchers from Virginia Tech have devised a low-input, high-throughput method for ChIP-seq that makes use of microfluidic technology in order to enable epigenomic profiling of clinical samples, where material is limited.
In a study published this week in Science Advances, the researchers demonstrated that the technique produced results that are comparable to standard ChIP-seq, using 100 cells or less. In addition, they applied it to cells from mouse brain, showing that it could identify epigenomic differences in cells from distinct brain regions. The study builds on previous work the group published in 2015 in Nature Methods, which also used microfluidics to for low-input ChIP-seq.
Chang Lu, senior author of the studies and a professor of chemical engineering at Virginia Tech, said the team is now working on applying the method to clinical samples, with a particular focus on brain cancer samples. He added that although he does not have specific plans to commercialize the method, it is something the university is considering. The university filed a patent related to the previous microfluidic-based method.
The main advance the researchers made from the 2015 version of the method is that they no longer rely on antibody-coated magnetic beads, which made it difficult to scale up the process and multiplex. Instead, the method involves functionalizing the channel surfaces of a microfluidic device by coating them with antibodies. The antibodies are held in place by DNA oligo linkers. Then the chromatin immunoprecipitation reaction can take place within the microfluidic device, and afterwards, the DNA can be sequenced.
In the recent study, the researchers focused on analyzing neuronal and glial cells from the mouse brain and demonstrated that they could run up to eight assays per microfluidic chip. In addition, they showed that the method, dubbed SurfaceChIP-seq, produced good results for three different histone modifications.
Andrew Adey, an assistant professor at Oregon Health and Sciences University who was not affiliated with the research but works on single-cell sequencing methods, said that the technique was interesting and definitely helped to move the field forward. "There's been a push to decrease the input for ChIP-seq for a very long time," he said, with limited success.
For instance, researchers from Harvard University and the Broad Institute modified their Drop-seq protocol to do single-cell ChIP-seq. While that technique does get down to the single-cell level, it does not result in many reads per cell — between 500 and 10,000. The SurfaceChIP-seq method, by contrast, generated 3 million and 13 million reads for 30 and 100 cells, respectively.
In the recent study, the researchers first tested the protocol on a cell line, using inputs of 30 cells and 100 cells, and looked at the histone markers H3K4me3 and H3K27me3.
Lu noted that the 30-cell input was sufficient to generate high-quality data when looking at H3K4me3, but because the signal for the H3K27me3 histone mark is not as strong, an input of 100 cells produced better results.
Next, they created microfluidic devices with four and eight channels and demonstrated that they could use the channels to either run different samples or to functionalize them with different antibodies to analyze different histone marks for the same sample.
Adey said that the ability to multiplex to analyze multiple features was an advantage of the method. With standard ChIP-seq, he said, the large input requirement is compounded for researchers who want to analyze multiple histone modifications.
Finally, the researchers used the method to analyze three different histone marks of neuronal and glial cells from the prefrontal cortex and the cerebellum of a mouse brain. They found that the cell types had different histone mark profiles depending on the brain region they were isolated from. Also, when they compared the results to RNA-seq data, they found that genes differentially expressed between cell types often involved different histone marking in the promoter regions.
Lu said that the team plans to apply the method to different sample types and has a particular interest in the "epigenomic dynamics involved in things like cancer development."
In addition, he said, the researchers plan to continue to improve the method. Adey noted that although the protocol was relatively straightforward, it would be limited to the few labs that have microfluidics capabilities. "In order to get to the masses, it would have to be commercialized," he said, adding that the platform is very amenable to that.
Lu said that the lab wants to improve the method to make it both easier to use and to decrease the input requirements even further, with an eye toward eventually enabling single-cell analysis. Right now, he said, the sweet spot is 30 cells. That yields 3 million reads, "which is high-quality data that's comparable to bulk data," Lu said. "I think our application has the potential to go down to single cell, and that's one aspect we're continuing to work on," he said.