A team from the UK and Denmark has developed a nanofluidic lab-on-a-chip design that makes it possible to do structural variant mapping, FISH analyses, and next-generation sequencing on the same DNA molecule.
"What we wanted was an integrated approach, where we get long-range information and we get short-range information," Kalim Mir, a visiting fellow in genetics at Harvard Medical School's Wyss Institute for Biologically Inspired Engineering, told In Sequence.
Mir, who was based at the University of Oxford's Wellcome Trust Centre for Human Genetics when the research was performed, co-led a study published recently in the Proceedings of the National Academy of Science illustrating that the chip could be used to find structural variants — including a large inversion and a large insertion — in DNA molecules from a previously sequenced African genome called NA18507.
In the same study, researchers demonstrated the feasibility of performing structural variant analyses, FISH, and high-throughput sequencing on the same strand of DNA — a combination that should make it possible to get an integrated view of a relatively large chromosomal region while maintaining haplotype phase information.
Consequently, the group is continuing to improve and automate the technique, which may one day prove useful for not only structural variant analyses in general but also for assembling genome sequences.
"Ultimately, we think these kinds of mapping approaches could help to reconstruct short-range sequence information and then assemble that sequence," Mir said, noting that this might "edge us closer towards de novo sequence assembly."
If and when single-molecule sequencing platforms emerge for generating reads that stretch to a million bases or longer, he added, it may be possible to get long-range structure information without this sort of chip-based mapping method.
But in the meantime, Mir predicted that there will be an advantage to having information afforded by such mapping approaches, which provide an intermediate scale between the chromosomal view offered by FISH and the sequence-level data coming out of short-read sequence data.
"If, in the future, we've got sequencing approaches that themselves give you long-range information then we wouldn't need this approach," Mir said. "But if your sequence read is not a megabase then it could be useful."
Along with an ongoing interest in understanding variation at the DNA sequence level and its role in health and disease, researchers are increasingly keen to get a handle on structural variation in the genome — from copy number changes to inversions to sequences that are lopped out of one part of the genome and land in another.
Detecting the full suite of structural variants within a given genome can be tricky, though several strategies have been developed or proposed for finding such variants. For instance, groups such as OpGen rely on restriction enzyme-based methods to do high-resolution optical mapping of genomes — an approach that that company is bringing to a range of structural variant analyses and genome assembly applications (IS 2/19/2013).
Another firm, BioNano Genomics (formerly known as BioNanomatrix), is tackling the genome structure problem using high-throughput chips that stretch DNA in nanochannels. By enzymatically labeling specific sequence motifs in the DNA and imaging these molecules in the chip, it's possible to unearth structural information.
The technique presented by Mir and his colleagues in PNAS involves DNA stretching, too, though the methods for achieving this stretch — and for assessing structural variants — differ.
Their approach involves a nanofluidic chip with cross-like channels in the center. DNA from whole chromosomes is first added to the chip. After DNA extraction, individual DNA molecules can enter one arm of the cross. There, the DNA molecule becomes stretched by hydrodynamic forces exerted by fluid flowing into the channels perpendicular to it, coaxing the genetic material to stretch to around 98 percent of its full length.
"It's this actual chip design that helps it stretch out more than other methods have achieved," Mir said. "Typically in fluidics you wouldn't achieve this sort of 98 percent stretching that we get."
As such, the chip does not require the same type of precise manufacturing technologies needed to make chips with very narrow, nanoscale-level channels, he added. Instead, the newly described chips can be produced using relatively simple and inexpensive methods such as the injection molding technology used in plastics manufacturing.
"That kind of manufacturing can be very cheap and highly industrialized," Mir explained. "So I think we've got some advantages in terms of the fabrication of these chips."
Mir explained that once the DNA molecule is stretched within the relatively wide channel, virtually any of the existing methods could be used to map structural variation in it.
At the moment, the researchers are using a denaturation mapping approach, also known as melting mapping, which starts with simple DNA staining.
When DNA is stained with a standard, intercalating fluorescent dye, Mir explained, the entire molecule will fluoresce. But when the DNA gets denatured, this dye gets lost from certain parts of the genome more quickly than others.
In particular, sequences that are rich in adenine and thymine bases tend to lose the dye first, since the AT-bases form weaker pairs than their guanine and cytosine counterparts. So by using light and dark patterns along a piece of DNA within the chip — which can span a million bases or more — researchers can track ATGC areas on the molecule and map them back to a reference genome sequence.
"We can kind of get a signature of the genome," Mir said.
"With our mapping approach in the chip we're looking at molecules that are one to two megabases in length," he added. "So we can see structural variant information at that scale, which is kind of an intermediate scale where it's been hard to get information on structural variation."
In general, it's possible to tweak the melting temperature of DNA in a system using certain buffer ingredients. For instance, higher salt concentrations tend to stabilize DNA, hiking up the temperature needed to denature it, while the addition of a chemical called formamide lowers this melting temperature.
For the current analysis, the researchers didn't bother adding formamide. But they did include a renaturation step to make DNA more robust during the stretching and mapping step.
Following these denaturation and renaturation steps — achieved by heating the chip and then cooling it in an ice bath — the DNA molecules have returned to their original double helix shape, Mir explained, but they still show the light and dark pattern associated with denaturation, since the dye has been lost from AT-rich sequences.
For instance, when the researchers did denaturation-restriction analyses on more than 50 DNA molecules from the NA18507 genome and randomly selected 21 of these for further analyses, they were able to match all of the molecules back to the human reference genome.
From that data, the team dug up a large chromosome 16 inversion (previously detected in the genome by paired fosmid end sequencing), along with an insertion involving a stretch of chromosome 9 that had been translocated into chromosome 19. The method unearthed smaller structural variants in the NA18507 genome as well.
The group faced a bit more of a challenge applying the approach to the Jurkat cell line, which was originally developed from an acute T-cell leukemia sample. Those complications appear to be a consequence of cancer-related rearrangements in the genome, Mir noted.
Even so, the method provided some new insights about the Jurkat genome structure as well. For instance, Mir explained, previous FISH-based analyses have uncovered large cytogenetic changes to the Jurkat genome. But the new analysis also pointed to sequence scrambling at the sub-megabase level.
In their subsequent experiments, meanwhile, the researchers demonstrated that it's possible to analyze the same DNA molecule in multiple ways using their nanofluidic chip design. For those proof-of-principle experiments, they rescued a denaturation-renaturation mapped DNA molecule from the chip, amplified the DNA, and funneled it into FISH and/or next-generation sequencing analyses.
"With the sequencing reads, what we've basically done is reconciled different layers of information," Mir noted.
The location of mapped DNA molecules tended to coincide using the FISH and next-generation sequence data, supporting the idea that both methods can accurately pinpoint the source of a chip-extracted DNA molecule in the genome.
The ability to accurately interpret patterns from denaturation-renaturation mapped molecules depends on an accurate and complete reference genome, Mir explained, meaning it will be difficult to assess structure in parts of the genome where reference sequences are missing.
He cautioned that there are also certain parts of the genome that may not give a clear pattern, such as long stretches of especially AT-rich or GC-rich sequence.
On the sequencing side, the approach is somewhat limited by the multiple displacement amplification method used to amplify DNA that's been rescued from the chip, Mir noted, since this method is prone to contamination and other complications.
"If all our reads were human we'd need a lower number of reads," he explained. "But because a large proportion of our reads were from contamination, we had to have a reasonably high coverage."
For their PNAS study, the researchers sequenced the amplified DNA using the Illumina GAIIx. But the approach is expected to be compatible with any of the existing high-throughput technologies.
The version of the chips that were used for the current study were fabricated using relatively expensive lithography and reactive ion etching processes, though Mir explained that the same chip design should be compatible with simpler plastics manufacturing techniques — a switch that would make it possible to produce the chips for just pennies apiece.
Going forward, the chip is going to be used as part of a project known as Cell-O-Matic, which is focused on characterizing cancer cells.
For that project, researchers will use the chip for DNA extraction prior to sequencing, in some cases. For other experiments, they will perform both DNA extraction and mapping on the chip prior to sequencing.
Over the course of that project, Mir and his colleague hope to continue refining and automating the chip-based integrated mapping and sequencing methods, with an eye to potentially coming up with a commercial product in the future.
There is also interest in trying to spool longer DNA molecules through the chip — perhaps up to whole chromosome lengths — and in exploring other DNA amplification methods that are less fraught with complication than MDA.