September 27, 2011 | Just ten weeks after leaving Life Technologies, Kevin McKernan’s new joint—a start-up company called Medicinal Genomics—is gaining attention by sequencing the cannabis genome, which could have important implications for cancer treatment.
As Bio•IT World reported in 2007, McKernan was the co-developer of the SOLiD sequencing technology, acquired by Life Technologies (see, “The Drive for the $1,000 Genome,” Bio•IT World, May 2007). McKernan and his colleague Alan Blanchard actually played a key role in the purchase of Ion Torrent. The two Life scientists watched Jonathan Rothberg unveil Ion Torrent’s Personal Genome Machine at a major conference in February 2010 (Advances in Genome Biology and Technology).
“Alan and I put in a bid internally to consider [Ion Torrent] for purchase. We helped snowball that,” says McKernan. Many other executives got involved, including Kip Miller and Mark Gardner, who “tore apart the physics and circuitry” (see “The Ion Inquisition”). McKernan then served as the head of R&D under Rothberg, helping to integrate the Ion Torrent and the SOLiD sequencing teams for nearly 12 months, until he decided to leave last June.
One of the challenges running the SOLiD program, says McKernan, is that it went through multiple owners – first Agencourt, then Applied Bio, then Life Technologies. “There was never one general on it,” says McKernan from his home on Boston’s north shore. He didn’t want the Ion integration to turn into an ‘us vs them’ rivalry. “There are a lot of synergies between the platforms. It was a fun experience,” he says. Ion’s R&D team is “a force to be reckoned with now—they have a lot of people from 454, SOLiD, Helicos—it’s a really tight knit group.” After 12 months, McKernan felt it best to move on. “There needed to be one general and I just had to get out of the way,” he says. “That’s the healthiest thing from a technology point of view. It was running at too fast a clip.”
Much of McKernan’s focus at Life Technologies—and a key application of the SOLiD platform—was in cancer research. Although he is no clinical oncologist, it was his interest in the cancer-fighting properties of cannabinoids that sparked his new venture.
“I decided to leave Life because I started to get a bit more serious about it,” he says. “It was probably best to keep it clean and not have any overlap. [Life Technologies] is not in this business, but it’s never good to have these conflicts.”
McKernan’s scientific interest in cannabis dates back to 2003, when friends battling cancer started handing him papers describing anti-cancer properties of cannabinoids. His cynical view shifted as he read papers by scientists such as Spain’s Manuel Guzmán. “It isn’t all Cheech and Chong,” he says. For example, “there’s real data on cannabinoids shrinking tumors in rats in nine different tumor types including glioblastoma.” There is also good evidence that the compounds can prevent tumors from metastasizing.
The active anti-cancer cannabinoid is not tetrahydrocannabinol (THC), which gives marijuana its recreational lift, but the closely related, non-psychoactive cannabidiol (CBD). Sativex, a combination of THC and CBD, is manufactured by the UK’s GW Pharmaceuticals and marketed for pain management in patients with multiple sclerosis.
“That peaked my interest,” says McKernan. “Why aren’t they synthesizing those two compounds? That would be cheaper, you’d think.” But given these plants feature dozens of cannabinoids, it would be a very complicated synthesis. “That’s when I realized someone’s got to sequence these [genomes]. Cannabinoid expression is strictly controlled by genetics—if you sequence the DNA, you can predict how much THC and CBD the plant will have at an early stage.”
Whether the solution is synthetic biology or more traditional plant breeding to maximize cannabinoid levels, it is necessary to understand which genes are driving cannabinoid expression. McKernan would like to replicate the Amyris story—the successful synthesis (and distribution) of the anti-malarial compound artemisinin. “We’re not going to figure out how to optimize this without tearing apart those pathways,” he says.
Even though he only needed microgram quantities of marijuana for sequencing, McKernan wasn’t about to risk the wrath of Massachusetts law enforcement. “We’re doing everything we can to keep this as legal as possible,” McKernan laughs. The state’s threshold for criminal possession of marijuana is 28 grams, but McKernan decided to set up shop in (not surprisingly) the Netherlands. “It does happen to be the world’s largest plant registry. They take their botany and genetics very seriously,” he says.
The sequencing, ironically perhaps, was done by Life Technologies competitor, Roche/454. “That might create some controversy, but to be frank, if you’re looking to do a de novo plant genome, 750-basepair reads on that platform are the way to go,” says McKernan. Two individual 454 reads are almost sufficient to span the entire THC synthase gene, for example. He also gathered an “ungodly” amount—320x—of Illumina HiSeq data, supplied by McKernan’s former Agencourt colleagues at Beckman Coulter Genomics. That will eventually prove useful, but in the short term, there isn’t enough phasing information to perform assemblies on such highly polymorphic genes. “We’re going to combine the two [datasets]; they’re going to complement each other without a doubt,” says McKernan.
McKernan has put the Illumina data on Cannabis sativa into the cloud, while the assembly statistics are on the Medicinal Genomics website. He is assembling millions of 454 reads using software from CLCbio in his house (although in addition to being extremely polymorphic, the cannabis genome is very AT-rich, which hampers genome assembly). Likewise, 454 scientists will be assembling the data with their own program, Newbler. Those data will also be released in due course.
McKernan will follow the initial reference for Cannabis sativa with a reference for C. indica and other strains. But he’s running a business, so not all of those data will be released. “When we start finding variants that matter functionally, we’ll hopefully be doing that for other partners that value that data,” he says. “There’s so much to be done on this plant, we’ll probably remain focused on this until we’ve cracked it.”
For now, McKernan wants to encourage other groups to get involved. Another goal is “to incubate the assembly algorithms that are all somewhat hyper-focused on different organisms. I don’t know of [algorithms] designed to handle very highly polymorphic plant genomes, purely with next-gen data. We’re not set up to be a clinical company or apply anything to patients—we’re at the ground level to build a database that can guide this space. We’re trying to keep the focus on informatics—figure out the pathways and find genetic fingerprints that help the industries working in this space regulate it and monitor what they’re growing.”
The sequencing will stay outsourced. “There’s no reason for us to build that [capability] today. We can keep our expertise on the interpretation of the data and how that can be applied to cancer patients.”
Meanwhile, McKernan can observe the ongoing NGS market battle from a safe perch. “I think so much of our field has hyperfocused on who is going to be the sole winner. But the market is getting so big now. Is there a single winner in the computer industry? Is it Google or Cisco or Apple? This [NGS] field is going to be like that soon. All of these companies are going to thrive—they’re sitting on an enormous tide that’s rising... They’re all going to find a customer.” •
As the co-developer and champion of SOLiD sequencing, what got Kevin McKernan and his colleagues so enthused about Ion Torrent? “What got us excited was the potential to break the diffraction limit,” says McKernan.
What is the diffraction limit? It is difficult to optically resolve two features that are shorter than the wavelength of light used to measure them. Most DNA measurement systems use light wavelengths of 500-700 nanometers (because smaller wavelengths mutate DNA). But measuring closely packed DNA features smaller than 700 nm makes the two features blur together. In DNA sequencing, the beads need to be spaced 700 nm apart so the photon wavelengths can resolve them, which is highly inefficient.
“[In 2010] We were in the process of mapping out another version of SOLiD that would have multiple cameras,” says McKernan. “We always knew that that would eventually hit the diffraction limit, and we’d be fighting a brick fight with photons with the other guys [Illumina] who make the same kind of box.”
McKernan says the Ion Torrent technology has the potential to read features less than 700 nm. “That’s what opened our eyes—something as small as a transistor can get down to 30 nm. They’re not there today, but it has a very long runway. Whereas we felt that if we built this next [SOLiD] optical system with multiple cameras and so on, that would probably be the end of the road… We were all racing to that endpoint—even though it would be a few years out—and then what? Nanopores break diffraction limits but there are complications.”
On the other hand, the Ion platform not only breaks diffraction limits but also uses the same front end methods as SOLiD. The Ion Torrent platform also features a very different cycle time—what McKernan calls “real time detection” on nucleotide incorporation. Taken together, it is easy to see why McKernan and colleagues took the plunge.
Taking McKernan’s place on the SOLiD team is Michael McKenna, who formerly worked for Rothberg at Curagen, working on its own DNA ultra-thin gel sequencer. “He seemed like the perfect fit to shepherd this along,” says McKernan. But he deflected inquiries about Life Technologies’ other NGS program, a single-molecule sequencing system under the direction of Joseph Beecham that has been dubbed “StarLite.”
“I was always focused on SOLiD,” he says diplomatically. “It was very exciting work, I love Joe’s work. It’s fascinating—there’s nothing cooler than watching single molecules in real time.” K.D.