In securing the rights to a “third generation” sequencing technology years before it will be ready for market, Sequenom has laid long-term plans for potential diagnostic applications of sequencing, and has also expanded its genotyping toolkit beyond the fine-mapping capabilities of its flagship MassArray technology.
Last week, the company said it had exclusively licensed a single-molecule DNA sequencing technology from Harvard University. The technology is based on an optical readout method that is coupled with nanopore arrays. Amit Meller conceived of the technology, which is still at the proof-of-principle stage, while at Harvard. He has since joined Boston University’s department of biomedical engineering as an associate professor.
Sequenom also plans to license additional intellectual property that Meller’s lab has been developing at BU, and is still looking for other technologies that complement the nanopore approach.
“We have been looking at all of the new sequencing technologies as they emerge in the public domain,” Charles Cantor, Sequenom’s chief scientific officer, told In Sequence last week. In addition, like Meller, he is a professor of biomedical engineering at BU. But since the company’s long-term interest is in diagnostics, that placed a number of constraints on its choice, he said: sample prep had to be simple, the platform had to cater to projects small and large, and it had to be able to reach a low enough price to be practical for diagnostic use.
“It’s not yet clear whether in diagnostics, you are going to want to sequence whole genomes or selected parts of genomes or just individual genes,” Cantor said. “Clearly, you want to have one platform that scales.”
Also, judging from projects that have used current high-throughput sequencing technologies, “it doesn’t look like they are going to be able to scale to the cost structure that would be needed to make a commercial diagnostic product successful,” he said.
The cost of a diagnostic test, he added, must not exceed $100 to make it commercially viable, whether what it delivers is a single gene or an entire genome.
But the company sees potential for the technology even in the shorter term. Prior to diagnostic applications, a sequencing platform would give Sequenom an entry point into the growing whole-genome genotyping market. Unlike microarrays, Sequenom’s MassArray technology, which analyzes DNA by mass spectrometry, does not have a high enough throughput for whole-genome scans.
“What we are missing here … at Sequenom is something that is genome wide,” Cantor said. “And we did not want to go to any of the array technologies, because our experience in trying to get very high-quality data with samples on surfaces has not been terribly rewarding.”
Whole-genome scans using arrays, he noted, create a considerable number of false positive results. Many researchers currently rely on the accuracy and precision of MassArray to follow up on array results, an application that Cantor said is actually Sequenom’s “major business today.”
Sequencing, he believes, could replace array-based methods for genotyping or gene expression analysis. “If sequencing becomes cheap enough, it will substitute for the arrays, with the advantage that the data quality is likely to be higher, and therefore, false positives are likely to be lower,” Cantor said.
Meller’s technology, in particular, is a good fit with Sequenom, Cantor believes. He said he first met Meller “during the very early days of nanopore work” when Meller was a postdoc in Daniel Branton’s lab at Harvard, and “provided occasional advice on that project.”
Both the mass spectrometer and the nanopore array platform involve preparing samples and running biochemical reactions offline, prior to putting the samples on the arrays.
“We have shown that the kind of biochemical expertise that we have amassed here, the particular way that we like to design and format assays, that that will fit nicely,” Cantor said.
Also, both the mass spectrometer and the nanopore are “very high-content detectors,” he added. “These high-content detectors require, in the real world, very fast signal processing in order to be able to use them efficiently,” meaning analysis in real time, Cantor noted. “And that’s exactly one of the concentrations of expertise that our software team has.”
Finally, nanopores, like mass specs, have the potential for becoming a “general, enabling platform across a wide range of applications,” he said. Sequenom’s mass spec-based platform has applications in genotyping, gene expression analysis, epigenetic analysis, and sequencing, focusing on small-scale projects with high data quality. “It’s our expectation that once matured, the nanopores will offer the same wide range of applications and high data quality,” Cantor said.
But that is likely to be several years away. In general, developing a new DNA sequencing platform takes about five years, Cantor said, though “there might be other, nearer-term applications that use the same detection platform but have shorter development cycles,” such as genotyping or gene expression, he said.
Also, Sequenom is not done shopping yet. While Meller’s technology, which will likely generate relatively short reads, will be applicable to many sequencing applications, it might not be sufficient for whole-genome sequencing that requires assembling short reads across difficult regions, Cantor said. “I am not convinced at this point that a single technology will enable very cost-effective whole-genome sequencing,” he said. “It may have to be done with multiple technologies.”
For that reason, he said, “we are still looking actively at the possibility of in-licensing other methods that would complement the nanopores,” and would help to assemble short reads into the correct haplotypes.
The nanopore platform is still “very, very young,” Cantor said. “There is a lot that has to be matured, and there is a lot that has to be learned.”
‘Feeding the Beast’
At the heart of the technology developed by Meller’s lab lie two principles: detecting optical signals, instead of electrical signals, from a nanopore; and using nanopores to strip off labeled probes attached to the DNA. This will result in “flashes of light that can give you information to directly sequence the DNA,” Meller told In Sequence last week (see also GenomeWeb News, In Sequence’s sister publication, 8/8/2006).
Last year, he won a three-year, $2.2 million grant from the National Human Genome Research Institute’s advanced sequencing technology program, following a smaller award in 2004.
Crucial to the process is a DNA conversion scheme that replaces each base in the original DNA with two oligonucleotides that come in two flavors: one type is labeled with a green fluorophore, the other with a red one. This two-bit encoding means that researchers do not need to read the DNA with single-base resolution but merely record “the flashes of light coming from the unzipping process of these oligos,” Meller said.
Meller has been collaborating with LingVitae, a Norwegian company founded by Preben Lexow in 2002, on a ligase-based DNA conversion process that uses pre-synthesized oligos. “They invented the conversion technology,” Meller said, noting that it has been improved significantly since the start of the collaboration. “We are focusing on the other end, which is the readout of the converted DNA.”
“It’s not yet clear whether in diagnostics, you are going to want to sequence whole genomes or selected parts of genomes or just individual genes. Clearly, you want to have one platform that scales.”
At present, Sequenom has no agreement with LingVitae, but Cantor did not rule out that the two companies would work together in the future.
Neither Meller nor Cantor would comment on the accuracy of the DNA conversion process. “It’s a big concern that we are aware of,” Meller said, noting that his lab has been working on ways to improve it, both with LingVitae and on its own.
“All I am going to be able to tell you at the moment is that there are alternate ways to feed this beast,” Cantor commented.
Over the last year, Meller and his colleagues have been working on shifting proof-of-principle studies from protein nanopores to solid-state nanopores. In particular, they have been able to construct arrays of up to 36 2-nanometer pores that are compatible with fluorescent readout.
This involved developing methods to form pores “much faster” and make them more uniform, and inventing a method to treat the pores so they would allow DNA to pass through them. “Now we are in really good shape where our yield, in terms of how many pores that we fabricate are really usable, is close to 100 percent,” Meller said.
Meller’s group also demonstrated this year that it is able to unzip the DNA in the solid-state pores and read up to two bases, or two bits, of DNA that was converted to labeled oligos.
“It’s the initial demonstration that takes a lot of time because you need to make a device, and a special microscope that can [read the signal],” Meller said. “Extending it to more and more [bases] should not be such a major effort.”
Besides extending the read length, Meller and his colleagues are now working on hardware and software to enable them to read many pores simultaneously. They are also hoping to increase the density of the pores to 100 per array, and ultimately 2,500 per array.
His goal is to show by the end of the NHGRI grant, which runs through 2009, that the technology can read DNA with a read length between 20 and 40 nucleotides “at a very low cost.”
Though he would not mention a specific price, he aims for a cost per base that is cheaper than current next-generation sequencing technologies. These technologies, he explained, cannot get much cheaper because “they need some key enzymes, and they need large quantities of those enzymes.” His group, on the other hand, has ideas for “how to minimize the use of enzyme” used in the DNA conversion process, “and that would drop down the price by a lot.”
Meller saw Sequenom as a good fit for helping to develop his technology into a product because the company has had a good track record, based on “the really remarkable work they did with the MassArray, which is a highly sophisticated instrument,” he said.
“They bring a really large expertise in this area that we think will be important when we will start to engineer our technology toward an instrument,” he said.
“Amit is a tremendously bright instrument developer and technology conceptualizer, but he is not an engineer, he is not a software developer, and he is not a biochemist,” Cantor said. “So you can see how our skill set is going to nicely complement his skill set.”
Though Sequenom has not decided on how it will contribute to the development of the technology, Cantor said it is likely the company will support research in Meller’s lab as well as conduct R&D in house.
Meller hopes that Sequenom’s expertise will come into play even before his lab has finished the proof-of-principle studies he has planned for the next two years, enabling them to have a product “very soon after that.”