BALTIMORE – Researchers from Cold Spring Harbor Laboratory have developed a new Cas9 target enrichment method for nanopore sequencing that promises to improve on-target coverage and read length.
The method, named Affinity-based Cas9-Mediated Enrichment (ACME), was built upon the existing nanopore Cas9 Targeted-Sequencing (nCATS) approach previously developed by Winston Timp's lab at Johns Hopkins University. By incorporating a background reduction step into nCATS, the researchers reported increased on-target sequencing depth and improved the end-to-end read length with the new method.
The nCATS method was the first effective approach to achieve targeted sequencing using nanopore long reads, said Shruti Iyer, a graduate student in the lab of W. Richard McCombie at Cold Spring Harbor Laboratory and the developer of ACME. However, because nCATS does not include a background reduction step, it produces a pool of both target and non-target DNA fragments, she said, setting off a competition for pore occupancy and decreasing the overall coverage of on-target reads.
ACME "is a modification and a slight improvement on the nCATS approach," said Iyer, who presented the new method, previously described in a preprint on BioRxiv, at this year's Advances in Genome Biology & Technology (AGBT) annual meeting earlier this month. "We felt that we needed something to reduce the background that gave our targets more of a chance to be sequenced."
Mechanistically, ACME leverages binding between the Cas9 nuclease, which contains a 6-histidine tag, and non-target DNA fragments after the cleavage step. By applying a His-Tag-specific cleanup step with magnetic beads, the researchers were able to pull down Cas9-bound non-target fragments from the sample, facilitating more target DNA to make it onto the flow cell.
The cleanup step has made "a huge difference," both in terms of sequencing depth and read length, Iyer said. For one, she said the on-target sequencing coverage has improved "across the board," with an increase anywhere from two- to tenfold.
In addition, she said, ACME enabled single reads spanning the target to be as long as 100 kb, up to three times the read lengths achieved by nCATS. The data also showed that even for targets 90 kb to 120 kb long, ACME helped capture two to 20 reads spanning the entire targets, whereas these counts were between zero and two reads without ACME.
"I've always been interested in trying to generate as long reads as possible for targets of interest," said Iyer, adding that longer reads can lead to larger tiles when covering large gene targets, potentially bringing down sequencing costs by requiring fewer Cas9 guides and making the mapping easier.
The researchers further tested ACME's ability to detect structural variants by comparing it with whole-genome sequencing data generated on Pacific Biosciences, Oxford Nanopore Technologies, and Illumina sequencers of SK-BR-3, a commonly studied HER2+ breast cancer cell line. The idea was to see if ACME was able to detect the SVs within the target regions that were previously identified within these whole-genome datasets, Iyer said.
The results showed that ACME detected all 15 SVs within the targets that were previously identified by PacBio and Oxford Nanopore whole-genome sequencing.
Similarly, the team compared ACME's capability to discern single nucleotide polymorphisms using the three SK-BR-3 whole-genome datasets. They found that ACME performed on par with whole-genome PacBio and Oxford Nanopore sequencing data when comparing the long-read platforms to Illumina, which is the gold standard for SNP detection.
However, Iyer said it is important to note that the initial benchmarking of ACME was conducted using 20 μg of what she called "pooled DNA libraries," which were combined from four parallel ACME preps. In real-world applications, especially with rare or tumor samples, it is unlikely to obtain 20 μg of DNA, she acknowledged.
To account for that, she also investigated the performance of ACME using a so-called single sample prep with 5 μg of DNA input. During the initial rounds of enrichment experiments, the reduced DNA input led to a drop in coverage and full-length reads, especially for some longer targets, she said.
However, after switching from old Cas9 chemistry to Oxford Nanopore's new standalone Cas9 kit that was released around late 2020, the team saw improved performance for ACME, with a sequencing coverage that was "very close to" what they used to achieve with pooled DNA samples, according to Iyer.
In terms of SVs, Iyer said that, using 5 μg of DNA and the old Cas9 chemistry, the team was able to detect 12 out of the total 15 SVs that were found by PacBio and Oxford Nanopore whole-genome sequencing. With the higher coverage delivered by the new standalone kit, Iyer said she is hoping to find the remaining three SVs with an even lower input DNA amount.
While ACME can help achieve a reduction in off-target background, it currently still has the trade-off of requiring a larger amount of starting DNA compared with nCATS, which needs around 3 μg of input DNA. "That's definitely something to keep in mind," Iyer noted, adding that the team is hoping to bring the DNA input down to within 3 μg.
ACME takes nCATS "a step further" by not only enriching targets of interest but also removing areas that are not targeted, said Timp, the developer of nCATS. The technology can be especially useful for researchers who hope to reduce off-target reads during targeted long-read sequencing, he added.
In addition to targeted ligation methods like nCATS and ACME, Oxford Nanopore recently released a so-called adaptive sampling feature, which achieves software-controlled enrichment by selectively sequencing regions of interest during a sequencing run.
Timp pointed out that while methods like nCATS and ACME are generally advantageous to achieve high coverage of one or two dozen target regions that are between 10 kb and 100 kb long, adaptive sampling may be more suitable when it comes to a large number of targets.
"Right now, if I had to do a lot of regions, I probably would resort quickly to using adaptive sequencing even with a lower coverage, because I just don't want to go through the rounds of design of [Cas9] guides," he said. There is still a need for the field to find better and easier ways to make guides that are effective, he added, while increasing the assays' multiplexing capability.
Iyer said ACME can currently target 11 regions within a DNA sample while maintaining good sequencing depth across all targets. She said the team is planning to expand the number of targets moving forward, noting that the challenge is to make sure the guides don't like conflict with one another.
In addition to expanding the number of targets, Iyer said another goal for the team is to achieve sequencing of multiple samples on the same flow cell. To accomplish that, the researchers are exploring the best way to make native barcodes compatible with the ACME chemistry.
"If you have more samples going on a flow cell, you're bringing down the cost of sequencing, as well," she said. "That's something we're actively working on."