From single molecules to Sanger, Kevin Ulmer has spent more than two decades seeking the ultimate sequencing solution.
By Kevin Davies
September 28, 2010 In 1980, Cambridge University biochemist Fred Sanger and Harvard physicist Walter Gilbert shared the Nobel Prize for chemistry for establishing the first generation of DNA sequencing technologies. Sanger became just the fourth scientist to earn a second all-expense-paid trip to Stockholm following his earlier win for protein sequencing. “It’s much more difficult to get the first prize than the second one,” Sanger said humbly. “I think I was lucky.”
Compared to the Sanger dideoxy method, Gilbert’s approach, developed with Allan Maxam, had the drawback that it involved some fairly noxious chemicals. Kevin Ulmer, then a postdoc at MIT, pragmatically preferred the Gilbert method because the two men were just a mile upriver at Harvard. “I remember the first time I got the can with the hydrazine in it in the lab, regaling my lab mates with how this was a hypergolic propellant for the NASA moon lander,” Ulmer chuckled.
Ulmer was interested even then in devising an automated DNA sequencing machine, and intensified after a stint in Japan, where he met physics professor Ashiyoshi Wada, who had put together a kiretsu of major corporations to attempt an industrial-scale Japanese genome project. Those efforts were overshadowed as Caltech’s Leroy Hood and Lloyd Smith built the first automated fluorescence sequencer in 1985, commercialized by Applied Biosystems (ABI).
Meanwhile, Gilbert, the co-founder of Biogen, decided to mount a private genome project even before the official launch of the Human Genome Project (HGP). “Wally being Wally, he was getting lots of media attention,” recalled Ulmer. “He simply wanted to build a factory in Southeast Asia, fill it up with automated Maxam-Gilbert sequencers… and just go at it in a commercial environment. It was the Celera business plan a decade earlier.”
Gilbert’s company, Genome Corp., never came into existence after the stock market crashed in October 1987, but he sensed the broader implications of the HGP. In 1991, Gilbert presciently published an article in Nature urging biologists to become computer literate in order to manage and interpret the human genome.
In 1987, Ulmer also founded his first biotech company, called simply SEQ (pronounced “seek”). He set up shop on the second floor of his house overlooking Cohasset Harbor on Boston’s South Shore, until his wife relegated him to the unheated basement. “I’m trying to sit there with my fingerless gloves and a cup of coffee. You could see my breath talking on the phone,” he recalled.
With a background in biology, physics and engineering, Ulmer sought to devise a new method of sequencing single molecules of DNA. He considered using newly introduced electron microscopy techniques, but knew there would be challenges of sample preparation. Instead he proposed threading them through a tunneling gap in a sub-microscopic nanopore, but after discussions with experts at Cornell (where the Pacific Biosciences technology was later developed) he realized that the technology was not up to the task. Of one thing, however, he was certain: DNA sequencing was an information service business. The original SEQ business plan in 1987 discussed sequencing a human genome at a price that medical insurance would reimburse. “For some reason, I picked $1,200 instead of $1,000,” he says, but to a close approximation, he was talking about “the $1,000 genome” in the 1980s.
In 1991, SEQ received its first major investment from Bob Johnston, a Princeton venture capitalist, with whom Ulmer first worked at Genex in 1979, and Richard Horan joined as CEO. Ulmer had settled on enzymology and a method he called single molecule sequencing by fluorescence (SMSF). There were similarities with an approach developed by the Los Alamos group of Richard Keller, which used an exonuclease enzyme to nibble off labeled bases one at a time from strands of DNA like a PacMan. Ulmer’s idea was to attach a bead to a single molecule of DNA, hold it in optical tweezers, then flow liquid so it unfurled like a flag in the wind. Then he would add exonuclease: if he could coax the cleaved, unlabeled nucleotides to fluoresce in a low temperature glass matrix, SEQ would have the makings of the first “next-generation” sequencing (NGS) system.
Ulmer persuaded Steven Chu, who would later win the Nobel Prize for his work with optical trapping and who is now President Obama’s Secretary of Energy, to join SEQ’s advisory board. He spent many weeks doing experiments in Chu’s lab, befriending a talented postdoc, Steven Quake, who would later become a co-founder of Helicos Biosciences. Ulmer set up SEQ’s first dedicated laboratories at the Sarnoff Corporation in New Jersey, and by 1996 had recruited a pair of single-molecule experts from AT&T Bell Labs—Jay Trautman and Tim Harris.
Progress remained slower than expected, however. In 1997, the same year that his patent on “Methods and apparatus for DNA sequencing” issued, Ulmer resigned. He’d been commuting to Princeton for six years and was disillusioned by some of the directions the company had taken. Although Johnston considered Ulmer something of a Renaissance Man, he felt the SEQ founder had underestimated the technological challenges. “Like a lot of other visionaries, he can look into the future,” Johnston was quoted at the time. “But often that means you focus on the mountain three or four mountains away and don’t realize how tough it’s going to be to get through all the valleys.”
A further frustration was money. A new wave of genomics start-ups, including Human Genome Sciences, Incyte Genomics, and Millennium Pharmaceuticals, were garnering lavish amounts of funding, promising to turn genes into drugs. “We were sort of three-day-old fish, shopworn, and the technology was too complicated,” Ulmer admits. “We were never able to catch that same wave.” Even an effort to coax Eric Lander (now the director of the Broad Institute) to join the SEQ board fell through when Lander co-founded Millennium instead.
Horan tried to push SEQ as a DNA sequencing service using ABI instruments on a large scale. “As Bloomberg is to financial services, we hope to be to genomics,” he said in 1997—a theme Craig Venter and Celera would soon echo. But SEQ didn’t have the money to pull it off. Tim Harris successfully shifted the firm’s focus to high-throughput microscopy for drug screening, before Amersham acquired the company in 2000.
Back in Cohasset, Ulmer had a new idea to build “a genomics company for the rest of us,” focusing not on disease but potentially lucrative cosmetic traits. He incorporated the company as Vanity Biotech, complete with a peacock feather logo, but when his venture partner rejected the name, Ulmer changed it to Pavonis (‘peacock’ in Latin). The initial plan was to catalog the active genes in human hair follicles, but his lab had only 1% of Celera’s assets. “Venter had 300 ABI 3700s, I had three. He had a $75-million supercomputer, I had a $750,000 computer.”
Ulmer’s timing was off again: One of his venture backers joined the dot com boom instead, and when the stock market crashed in the subsequent bust, bankers seized the company’s accounts. Pavonis was grounded for good.
Ironically, the last biotech company to go public successfully before the IPO window closed was a cancer diagnostics company called EXACT Sciences, the latest brainchild of Stanley Lapidus. Ulmer had helped Lapidus write the first patents for EXACT on detecting cancer DNA mutations in stool samples and later consulted for EXACT to develop “digital PCR” for single-molecule amplification. A couple of years later, Lapidus and Ulmer joined forces again, after Lapidus decided to build a company, Helicos Biosciences, based on Stephen Quake’s single-molecule sequencing technology. Ulmer gave Quake’s technology a ringing endorsement—he knew Quake well and had even been invited to his wedding.
Ulmer helped Lapidus hire some superbly talented scientists and engineers, including Tim Harris as director of sequencing technology. “I told Stan three things,” Ulmer recalled: “a) ‘I believe we’ll make the technology work,’ which they’ve done. b) ‘I reserve judgment as to whether you’ll be a commercial success. And c) ‘You’ve got the wrong business model,’ because they were still chasing the idea of selling an instrument and reagents.”
TWENTY years after SEQ, Ulmer was ready to pursue his latest vision for next-generation sequencing. “You keep pushing the boulder up the hill and hope that Sisyphus eventually gets there,” he said, with that inordinate fondness many life scientists have for Greek mythology. After two decades trying to reinvent DNA sequencing, Ulmer had surprisingly come full circle. The title of his keynote lecture at a CHI conference in 2007 said it all: “Sanger sequencing is dead: Long live Sanger sequencing.”
While Ulmer welcomed the arrival of NGS for its “wonderful infusion of energy, interest and capital” to an “almost moribund” space, he felt that all of the early commercial platforms—454, Illumina, ABI and Helicos—had their problems. His most detailed criticisms were reserved for Helicos, a technology he knew intimately. Indeed, his frank assessment of the company’s limitations—pricing, read length, optics and more—drew a swift rebuke from the Helicos’ lawyers.
But Ulmer didn’t want to talk about the future as much as revisit the past. He had been born again. After reviewing the flaws in NGS technology, Ulmer had come to the conclusion that there was nothing fundamentally wrong with Sanger sequencing. “It needs some polishing and updating, but we believe it can outperform all of these next-gen methods by a substantial margin,” he said. Ulmer reckoned a reformulated Sanger production line could outperform all the other NGS technologies. “We have a 30-year global investment history with Sanger. We understand it intimately, its nuances and subtleties, we know how to tweak it and modify it. We understand the error properties and it gives the longest reads obviously of any technology out there by a large margin.”
Ulmer had also returned to another early belief, namely that medical researchers were more interested in obtaining sequence information than running sequencing instruments. “Sequencing is an information services business. It’s not an instrument reagent business. Everyone wants sequence information, and they want it in vast quantities at very low prices.” Abandoning boxes instantly removed the physical constraints imposed by cramming lasers, lenses, robots and reagent bottles into a box, and enabled new economies of scale. DNA sequencing would become a commodity, “something like the printing press that produces your Sunday newspaper, not a room full of desktops.”
Ulmer announced the launch of his new company in July 2007. He phoned Gilbert, vacationing on Cape Cod, to ask if he could revive the name “Genome Corp.” Gilbert, who was retired and more interested in mounting amateur photography exhibitions, had no objections. After attracting $750,000 in seed funding, he leased lab space in Providence, RI, in the same building as NABsys (see page 56).
Ulmer hoped that potential investors would appreciate his track record to understand his somewhat curious reversion to Sanger sequencing, but his biggest technical challenge was the same as everyone else—imaging. How many pixels per second can be captured and processed that result in a base call? At a sub-microscopic scale, advanced optics don’t scale. The objective lens in the HeliScope, for example, had a narrow field of view of just 100x150 microns, requiring multiple frames be taken very rapidly. With Sanger, however, the bands of DNA are much bigger, so the optics do scale, as did the process of making giant electrophoresis films. Ulmer’s limitation wasn’t how fast he could run the gels, but “how fast can I image a football field?”
The initial plan was to start delivering high-quality DNA sequence in 2010. Ulmer fancied his chances of winning large NIH sequencing contracts and putting commercial sequencing firms out of business. “I want them buying sequencing the same way they buy toilet paper for the military,” said Ulmer. Those early purchase orders would fund the completion of his sequencing factory and the necessary IT infrastructure.
The long-term opportunity, however, was resequencing human genomes on a massive scale. “Ideally, that value chain would go all the way to the patient, who could use discretionary dollars, or medical insurance. That was a stated goal in the original plan for SEQ.”
“What would it take for Kaiser Permanente to take its entire membership and have them sequenced? It’s not just a question of price. Even if I could sequence them all today, they wouldn’t know what to do with it. They need to know: How does this save me money? How do I cut the cost of delivering health care by knowing all my members’ genomes? How does it allow me to customize their medical care in a way that has cost benefits?”
Unbeknownst to Ulmer, Silicon Valley-based Complete Genomics was also choosing a service model. When Ulmer visited Complete co-founder Rade Drmanac in San Francisco in July 2008, it was clear, “we were singing from the same hymnal,” not only about the advantages of a service model but also the desire to focus exclusively on human genomes. “It’s the only genome that makes sense to sequence millions and millions of times,” says Ulmer.
But Ulmer, who was working with just a single technician, didn’t realize that Drmanac was ready to start sequencing. Ulmer pressed on: he signed a term sheet with Bill Frezza at Adams Capital Management, who saw instant parallels to the Taiwan Semiconductor Manufacturing Corporation (TSMC), the world’s largest semiconductor foundry, which builds wafers for semiconductor clients. “That’s what Genome Corp. should be—the TSMC equivalent for genomics,” said Frezza. “You sequence genomes, but don’t chase upstream trying to do databases, diagnostics, and therapeutics. Allow the ecosystem to grow up around you.”
Frezza asked Tim Harris to handle the due diligence. Everything was positive, even their analysis of third-generation technologies such as Pacific Biosciences. But once Complete Genomics emerged from stealth mode in October 2008, Ulmer couldn’t assure Frezza that he could have his service ready and cheaper than Complete.
The global financial meltdown dealt Ulmer another body blow. Ulmer raised the white flag and called up Complete CEO Clifford Reid, who readily hired Ulmer as a consultant to advise on the scale up of Complete’s first genome center. “They had three years and a $35 million head start on us. If we’d started at the same time with similar amounts of funding, who knows which one would have wound up winning?”
As the field prepares for third-generation sequencing, Ulmer is still pondering his next move, and says he has several irons in the fire. (He is currently biding his time working on a project at the Woods Hole Oceanographic Institution.)
He could try another tack, however. He says philosophically: “If someone would just pay me not to start new ventures, the global economy wouldn’t collapse!” •
This article also appeared in the September-October 2010 issue of Bio-IT World Magazine. Subscriptions are free for qualifying individuals. Apply today.