Sugar is Sweet. . .
By Laura Stephenson Carter
Engineers build bridges and buildings, and scientists fiddle with test tubes and petri dishes, right? Well, not quite. Although there are philosophical differences in the way engineers and scientists approach their work, the work they doand the tools they use are often exactly the same nowadays.
A case in point is a biotech start-up company based in Lebanon, N.H., that is revolutionizing the way therapeutic proteins are made. It's a privatelyfunded commercial enterprise, but it wouldn't exist if it weren't for expertise from both Dartmouth Medical School and Dartmouth's Thayer School of Engineering. Even more to the point, it wouldn't exist were it not for the ways that engineers and scientists have pooled their skills.
Tillman Gerngross, left, and Charles Hutchinson,
rightboth members of Dartmouth's
engineering facultyfounded a biotech startup
that has drawn on expertise from several
Dartmouth Medical School faculty members.
Pooling skills is nothing new for Dartmouth's medical and engineering schools. For years, faculty from the two institutions have collaborated in such areas as orthopaedic surgery, neurosurgery, and oncology. And since there's no animal lab at Dartmouth's engineering school, Jack Hoopes, D.V.M., Ph.D., the director of DMS's surgical research labs, is often, he explains, "the connection for a lot of projects that have involved engineering and animal studies."
As a result, Hoopes is among the Medical School faculty members who have been swept into the revolution in drug production taking place in the Upper Valley. Soon, he will be testing therapeutic proteins being made by the nascent biotech firm. He also plans to use their products to gain additional insight into how well animal models work to test such therapeutics.
The biotech company, GlycoFi, was the brainchild of a pair of Thayer School engineersTillman Gerngross, Ph.D., a biochemical engineer, and Charles Hutchinson, Ph.D., Thayer's dean emeritus. They have daringly engineered yeast to produce human therapeutic proteins. Their concept is daring because scientists had said it couldn't be done. Everyone "knew" that only mammalian cells most commonly Chinese hamster ovary (CHO) cellscould be turned into bioreactors capable of mass-producing therapeutic proteins.
But Gerngross has never shied away from questioning the scientific establishment. In 2000, he published an article in Scientific American titled "How Green are Green Plastics?" In it, he challenged what scientists had proposed as an environmentally- friendly way to produce plastic. They had developed a "green" alternative to manufacturing plastic in petrochemical factoriesgrowing it in plants instead. "I had the audacity to ask the question 'How much greener is it?'" says Gerngross. True, the plastic would be from a renewable resource and it would biodegrade, but his chemical engineering analysis revealed that there were hidden ecological costs; the biological breakdown of plastic releases the greenhouse gases carbon dioxide and methane, and fossil fuels would still be required to extract the plastic from the plants. "In fact," he claims, "the CO2 emissions are worse, land use is worse, water use is worse . . . and yet you save a little bit of petroleum. It was a counterintuitive finding that gave me some notoriety for a while."
And now Gerngross has achieved notoriety againthis time in the therapeutic protein-manufacturing industry. He marched in declaring that he'd found a way for the pharmaceutical industry to keep up with the soaring demand for therapeutic proteins. There are already more than 80 such proteins on the market and some 700 more in clinical development. But there isn't enough capacity in labor- intensive CHO-cell "factories" to produce anywhere near all the new drugs that will likely be approved over the next couple of years.
"The capacity to make [therapeutic proteins] worldwide is already consumed by the current drugs that are in the marketthese big blockbusters like Enbrel, Factor VIII, erythropoietin," explains Hutchinson. Enbrel is used to treat rheumatoid arthritis; Factor VIII facilitates blood-clotting in people with hemophilia A; and erythropoietin, which stimulates the production of red blood cells, is used to treat anemia. Since the year 2000, some 60 million people have benefited from using therapeutic proteins like these.
Within each of the body's billions of cells, there are 20,000 to 50,000 different genes, most of which are responsible for making proteins such as enzymes, hormones, and antibodies. Therapeutic proteins are proteins developed to treat certain diseases or conditions. For example, insulin is administered to people with diabetes to compensate for their lack of that protein.
More than half of all the drugs in development today are therapeutic proteins and antibodies; other kinds of drugs include small-molecule compounds, vaccines, and gene and cell therapies. Small-molecule drugs are generally taken as pills, while protein-based therapeutics are usually given by injection in order to prevent their breakdown in the gastrointestinal tract.
Modern biotechnology has its roots in the 1953 discovery of the structure of DNA. In the early 20th century, scientists had learned how to harvest small amounts of therapeutic proteins from tissues and urine. With but a few exceptions, however, the amounts produced were too small to be practical for therapeutic use. Insulin, one of the exceptions because of its enormous therapeutic value, was painstakingly isolated from pig pancreatic cells.
But by the 1980s, advances in genetic techniques made it possible for scientists to identify and isolate the genes that produced therapeutic proteins. In 1982, insulin became the first genetically engineered protein. And by the early 1990s, other therapeutic proteinserythropoietin, human growth hormone, and othershad been developed. Now the pharmaceutical industry regularly comes up with new biopharmaceuticalsproteins produced by genes introduced into cells.
Such drugs are now used in the treatment of cancer; HIV infection, AIDS, and AIDS-related syndromes; autoimmune disorders such as rheumatoid arthritis and systemic lupus erythematosus; infectious diseases; neurological disorders; and heart disease.
Today, most therapeutic proteins are mass-produced in huge fermenting vats filled with genetically engineered mammalian cell lines, usually CHO cells. Scientists are also developing alternative production methods, such as harvesting proteins that are produced by transgenic animals, plants, or insects.
"But," says Hutchinson, "we didn't think that was the right answer." The right answer, according to Gerngross, was to genetically engineer yeast cells so they would churn out human proteins. After all, yeast and fungi have been used for years to manufacture industrial proteinslike the enzymes used in laundry detergents and animal foods or used to make stonewashed jeans. So he figured it shouldn't be too hard to get yeast to make other kinds of proteins, even human therapeutic ones.
"These fungi could make boatloads of protein," he says. The underlying proteins are the same in humans and yeast. Only the sugars that adorn the proteins are different. "So we felt that, if we could fix that problem, we could combine the best of both worldsnamely robust industrial processing . . . with making exactly what you want."
Only mammalian cells naturally attach the right sequence of complex sugar molecules onto proteins, in a process known as glycosylation. And the process of glycosylation is different in lower organisms than in humans and other mammals.
GlycoFi's principalsincluding Tillman Gerngross (top photo)have drawn on lots of DMS expertise to get their company going, including (above) Jack Hoopes, right, director of the surgical research labs, as well as, from the left, Lionel Lewis, Sujatha Sundaram, and Rendy Strawbridge.
Although some human proteinssuch as insulin and growth hormone are not glycosylated and thus can be produced in bacterial hosts such as E. coli, about 70 percent of human proteins, including antibodies, are glycosylated. Therapeutic proteins that are not properly glycosylated are useless or even unsafe in the human bodythey might have no effect, they might be cleared too quickly to do any good, or they might even cause an immune response.
At first, Gerngross thought his idea was "brand spanking new." But he soon learned that he wasn't alone in his quest to get yeast to produce human-like proteins. Two respected research groupsone in Japan and one in Belgiumhad been trying to do so for more than 10 years but had made little progress. However, by thinking like an engineer rather than a scientist, Gerngross has been able to tame yeast to do the seemingly impossible. Using what are called combinatorial libraries, GlycoFi tested specific combinations of enzymatic genes from an array of more than 1,300 possibilities. As a result, they were able to do experiments much fasterat a rate of 900 a weekthan in a typical lab, where one experiment can take from a few weeks to a few months to complete.
Each reaction in the synthesis of a therapeutic protein is carried out by enzymes. "So the hard part is how do you line them all up?" says Gerngross. "The first enzyme does its thing, and then the next one does its thing, and the next, so that we have this assembly line. Right? You can't put the windscreen in until you have the car painted. You have to do these things in a sequence. You can't put the mirror on before you actually have doors on the car. So in a very similar way, you have to have a sequential approach to doing this kind of stuff. And the hard part is to line them up so that enzyme A is in this place, enzyme B is in this place, and so forth. That had not been solved.
"And so," he continues, "what we said was, 'We know what the enzymes are. We just don't know how to get . . . there.' So we separated the part of the gene that is involved in targeting from the part of the gene that has catalytic activity, and then we made combinations of these. Because we didn't know which one would take us to the right place, we combined them so many times with so many different things that we had a high probability of having at least one that would work."
Then the genes are inserted into the yeast cells, where they become active in the endoplasmic reticulumthat part of the cell, together with the Golgi apparatus, where newly made proteins get glycosylated.
"People will ask us, 'What do you do with the ones that don't work?'" says Hutchinson. "'Don't you go back and try to figure out why they didn't work?' And our answer is, 'No. We don't care. They didn't work. I mean, what else do you want to know about it?'"
Scientists want to understand why experiments do and don't work. Engineers want to solve problems. While science is "discovery driven," Gerngross says, "in engineering your metric of success is . . . solving the problem."
Scientists want to understand why experiments do and don't work. Engineers want to solve problems. While science is "discovery driven," Gerngross says, "in engineering your metric of success is . . . solving the problem. It's fascinating for me to see that we are building this enormous knowledge in the life sciences area, but we are pretty weak in translating that into . . . something that we humans will benefit from. . . . It takes engineering and it takes acknowledging uncertaintiesand engineering around uncertainty and making sure that what you're doing will capture all the uncertainty[to] have a high probability of overall success."
"This is not [meant] to be critical," adds Hutchinson. "Pure science is about truth and beauty. And engineering is about 'We've got a problem. We've got sick patients and we want to get proteins that we can give these patients to make them well.'"
During an interview with Dartmouth Medicine, Gerngross jumps up and strides over to a complicated chart hanging on the wall of the company's conference room. The chart (which is in the background of the photograph on page 46) is filled with rows of colorful diagrams depicting proteins in the assemblyline- like process of the human glycosylation pathway. Gerngross points out where various sugars are attached to the base protein.
The pathways for yeast and for human proteins are identical up to a point, he explains. "They do everything the same up to here," Gerngross says as he gestures at the chart. "Then yeast does some really funky stuff, which is not on this map."
Because of that "funky stuff," nobody had been able to get yeast to make human proteins before. "We were the first ones to do that," Gerngross says. "And we have moved them all the way"he points further along on the chart"to this stage here."
Hutchinson and Gerngross also have a much simpler, easier-to-understand version of this chart that they show investors and potential clients. Instead of rows of diagrams, it shows only six steps, which GlycoFi scientists refer to as "milestones."
"We are basically at step five, nibbling on step six," Hutchinson says. Even if they get no further, they've already gotten the yeast to produce a number of therapeutic proteins that are of interest to potential customers.
When Gerngross convinced Hutchinson that yeast could be engineered to produce therapeutic proteins, the pair was so sure that they had a commercially viable idea that they decided to set up a company rather than take the slower, academic approach of writing grant proposals and waiting for funding. Venture capitalists have provided over $7 million so far, enabling Hutchinson and Gerngross to create GlycoFi (the name is a contraction of "glycosylation fidelity"). In 2000, they set up shop in space at the Thayer School in Hanover. By early 2003, the company had grown to 30 people and moved to larger quarters in the Centerra Office Park in Lebanon. Their labs there are better equippedand more colorfulthan typical academic labs. They have an open floor plan, walls painted bright red and yellow, color- coded floors to demarcate different research and production sections, and lots of windows.
Ironicallyfor a company that prides itself on having taken a nonacademic route to get startedGlycoFi relies on numerous academics to keep going. Most of the GlycoFi scientists came directly from academic labs, some of them still have academic appointments at Dartmouth, and the company has even applied for academic funding for certain research projects on which they're collaborating with academic labs. GlycoFi scientists have also published their research findings in peer-reviewed academic journalsincluding a research report in a recent issue of the prestigious journal Science.
Charles Hutchinson (top photo)the emeritus dean of Dartmouth's engineering schoolis one of GlycoFi's founders. Among those whom he and his colleagues have recruited to their firm from DMS is Andy Stadheim (above), pictured setting up PCR reactions in stacks of yeast plates.
But GlycoFi is based solidly in the for-profit world, too. It has filed for patent protection on its products and has had a couple of rounds of venture funding and is planning another round soon. The company is racing to gear up to meet expected demand from the pharmaceutical industry. It aims to be a player in the multibillion-dollar therapeutic protein industry.
The firm's principals aren't especially worried about competition from other scientists who are developing alternate methods of producing therapeutic proteins. They expect, for instance, that there will be numerous regulatory issues associated with using transgenic animals or plants as drug factories. Transgenic organisms have been genetically transformed so they can produce human therapeutic proteins, which are then extracted and refined to produce drugs.
"The question in simple terms is how do you make a sterile goat farm?" says Hutchinson. "Or how do you make a sterile cornfield? We think the FDA [Food and Drug Administration] is going to have a lot of issues [with such production methods]whereas the approach that we have taken is . . . done in stainless steel reactors, with steam, water, in a confined area that you can sterilize."
Reports that Gryphon Therapeutics, a San Francisco biotech firm, has developed a synthetic version of erythropoietin, don't faze the pair, either. "They synthesized it, but they cannot put the sugars on the way we do," Gerngross declares.
In fact, GlycoFi's ability to control how the sugars are attached to the proteins suggests that they may also be able to make "designer drugs"drugs with specific molecular structures. Mammalian cells typically produce a spectrum of proteins, with sugars in varying places. But a "particular structure of glycosylation can have a very significant impact on the potency of the drug," says Hutchinson. "We clearly have, right now, the only tool that we have seen that allows the drug companies . . . to figure out which one of these glycosylation structures is the best one to use for a therapeutic."
GlycoFi hadn't expected to be able to make designer drugs. They had figured they'd be "the lowcost manufacturer," as Hutchinson puts it. "For generics, that's a huge issue," he explains, "because with generics you have to be the low-cost producer." GlycoFi had anticipated being able to turn out quantities of such proteins as the red-cell-boosting erythropoietin; interferon-beta, used to treat multiple sclerosis; alpha-antitrypsin, used to treat hereditary emphysema; and a variety of others. The line gets "really blurry whether you are a manufacturer [of generics] or whether you are someone that can make designer drugs," explains Gerngross.
But the company isn't in a position to sell anything quite yetgeneric or designer. When that day comes, what the firm sells could be a product but more likely will be a process. In other words, GlycoFi might manufacture drugs according to a client's specifications. But Gerngross and Hutchinson expect that typically clients will use GlycoFi's technologyfor a feeto make a certain protein. Hutchinson explains how they envision such client relationships unfolding: "We make a little of it. They look at it. If they like it, then they can license our technology, which would include the yeast strains for that protein only. And then they could take it and run with it."
Hutchinson explains how they envision their client relationships unfolding: "We make a little of it. They look at it. If they like it, then they can license our technology, which would include the yeast strains for that protein only. And then they could take it and run with it."
None of GlycoFi's work would be possible if it wasn't for recent advances in genetic engineering and in identifying the human as well as the yeast genome. "If you had set out to do what we're doing 15 years ago, it would have been harder," says Hutchinson. "The tools were not there. There's been a lot more science done of late which makes a lot of this possibleall this genome sequencing and so forth. We have the genome for the yeast. So we know everything about it."
"We could not have done this without a huge amount of man and woman years going into establishing the basic science," agrees Gerngross. "Without all these genes having been sequenced, we couldn't have done anything."
Still, GlycoFi is the outfit that has managed to put together all the pieces of the puzzle. Others had spent years studying various pieces individually without understanding the big picture. Then, says Gerngross, "here comes someone who's not even a player in the glycobiology field . . . [who] walks in and finds something that is relatively significant in importance to the overall medical community, and basically uses their information that they've generated to build something that is useful."
Gerngross laughs. "You can imagine that it's a very odd dynamic to be involved in."
Even other engineers at pharmaceutical companies have been surprised at GlycoFi's accomplishments. "When we did our second round of financing, [the investors] forced us to go talk to the potential customers, because they wanted to make sure that this made sense," says Hutchinson. "So we had to go talk to a large number of biotechs and biopharmas.
"It was fascinating," he relates. "We would go in to make these presentations and there would be 15 scientists from their R-and-D group sitting in the room. About five minutes into the presentation, you could see the lightbulbs going off. They were saying, 'Why didn't we think of this?'" Hutchinson chuckles. "It was really interesting to watch their reactions. Because immediately they saw, [once it was] laid out in a very structured kind of way, that it just made all kinds of sense. There's no reason why they couldn't have thought of it. But they didn't. They just flat out didn't."
Steve Hamilton (top) and Jim Cook (at left above)are among GlycoFi's DMS-trained staff members. Both of them work in the firm's strain development section. But Thomas Potgieter (below), who runs the fermentation room, did not bring a Dartmouth lineage with him to the company.
Gerngross thinks that the engineers at these firms "were just so used to solving the problems the way they used to solve them'optimize what we have and squeeze out another 5 percent'"that they had never considered taking the approach that GlycoFi had.
That approach promises to boost protein production well beyond current methods. Bioengineered yeast can make therapeutic proteins up to 20 times faster than CHO cells can, but the costs associated with yeast-based production are only a quarter to a third as much.
GlycoFi may have been founded by engineers, but many of its scientists are biologistsincluding several with ties to Dartmouth Medical School. For example, the company's strain-development sectionwhich is responsible for genetically engineering the different strains of yeast used in the GlycoFi process, such as Kluyveromyces lactis and Pichia pastorisincludes Stephen Hamilton, Ph.D., Juergen Nett, Ph.D., and T. Andrew Stadheim, Ph.D., all of whom did postdoctoral fellowships in Medical School labs, and James Cook, Ph.D., who used to work for a DMS immunologist.
Working at GlycoFi is not all that different from working in an academic lab, says Stadheim, though "we are, of course, focused toward a product." When he interviewed at GlycoFi before joining the group last year, he was intrigued. "What they were trying to do seemed so heroic," he says. "If we could succeed in doing that, then the impact it could have on the production of proteins for clinical use would be tremendous."
"It's been a lot of fun," he adds. "There's a lot of energy here [and] a sense of urgency, which is good. Everybody here understands that sense of urgency."
Research associate Teresa Mitchell is another former DMSer. She worked in Dartmouth physiology and microbiology labs for more than 20 years before joining GlycoFi last year. In the protein-characterization unit where she now works, she uses robots to help process the thousand-plus samples she is responsible for analyzing each week. Her unit analyzes the proteins after every step in the production process to determine whether they have been made correctly.
Margaret Worden, who was recently appointed vice president of clinical product development at GlycoFi, worked for DMS biochemist Constance Brinckerhoff, Ph.D., while earning a master's in engineering at Dartmouth in the mid-1980s. With more than 10 years of experience since then directing clinical manufacturing at other biotech firms, Worden will be lining up and managing collaborative projects with outside companies. GlycoFi recently announced a research collaboration with Baxter Healthcare Corporation, for example. "The science is even better than I thought," she says of the GlycoFi process.
Not all the scientists have Dartmouth connections, however. Thomas Potgieter, Ph.D., a process engineer from South Africa, runs the fermentation room. There, the genetically engineered yeast strains go through a fermentation process and secrete the desired proteins. He makes sure the conditions are optimalthe right mixture of media, the correct temperature, enough oxygen, etc.for protein production. He is perfecting the fermentation process on a small scale in preparation for ramping up to industrial-sized fermenters.
Working at GlycoFi is not all that different from working in an academic lab, says Stadheim, though "we are, of course, focused toward a product." When he interviewed at GlycoFi before joining the group last year, he was intrigued. "What they were trying to do seemed so heroic," he says.
The company has cultivated other DMS connections, too. Early in the firm's life, its founders consulted with Dartmouth immunologist Michael Fanger, Ph.D. In 1987, Fanger cofounded Medarex, Inc., a biopharmaceutical company that develops monoclonal antibodybased therapeutics to fight cancer and other diseases. "He understood immediately what we were trying to do," says Hutchinson.
"I think GlycoFi's system has the potential to be quite useful," Fanger comments. In fact, he adds, "GlycoFi has technology that Medarex could use."
Another DMS contact has been immunologist Randolph Noelle, PhD., with whom strain scientist Jim Cook used to work. "Randy has an ongoing interest in modulating the functionality of antibodies," says Gerngross. "There was a clear overlap between what we were doing and what he was doing. One of the first antibodies that we tried to make came out of Randy's lab. And, in fact, Jim Cook who at the time was still in Randy's labcloned the antibody.
"Jim Cook subsequently left Randy's lab and joined us," adds Gerngross with a laugh. "So there has been a fair amount of cross-fertilization."
Jack Hoopes, the head of DMS's surgical research labs, is another Medical School collaborator. "We have a joint grant between Jack Hoopes at the Medical School, Glycofi here, and my group at Thayer," explains Gerngross. "Jack has also been interested in some of the usage of erythropoietin in rats. He has this model where you look to see whether erythropoietin can be used to overcome some of the side effects associated with radiation, using animal models," says Gerngross. "We can make the drug," he explains, while "they have the animal model. And we would farm it out anyway, so why not farm it out right there."
Hoopes recounts an early conversation he had with Gerngross. "Tillman said basically, 'I've got these yeasts and I'm going to figure out this new way to produce protein. I need some help integrating this into an animal model, and I don't really know so much how to do that.'"
Hoopes, as it happens, was doing a study with DMS oncologist James Rigas, M.D., on using Epogena brand-name version of erythropoietin, or EPO for shortto fight the overwhelming fatigue experienced by patients undergoing combined chemotherapy and radiation therapy for lung cancer. Hoopes told Gerngross he was willing to test yeast-produced EPO in rats that were getting chemo and radiation therapy for lung cancer, and compare it to CHO-produced EPO.
Hoopes had another idea that he wanted to test, too, and he thought GlycoFi could help. He had always wondered whether all animal models were appropriate for testing human biopharmaceuticals. What if human EPO was less effective in rats than it was in humans. Hoopes wanted to compare a rat version of the EPO protein to the human version, so he asked GlycoFi to engineer the yeast to produce rat EPO for such a comparison. Hoopes and Gerngross have applied for a combined grant from the National Science Foundation and the National Institutes of Health to fund that project. "Our theory is that the human proteins work [in rats] but that they're not optimal in animals," says Hoopes.
Margaret Worden (top), a graduate of Dartmouth's Thayer School of Engineering, is vice president of clinical product development at GlycoFi. And Teresa Mitchell (above), who worked in Dartmouth Medical School labs for 20 years, oversees the protein analysis and purification process.
Hoopes could even discover that ailing pets might be better off if they are treated with therapeutic proteins specific to their species. For instance, human EPO is now used to treat kidney failure in dogs. But maybe dog EPO would be more effective. "This opens a huge box," says Hoopes.
He anticipates that Gerngross will want to test other yeast-produced proteins, too. "Because of our animal-model expertise in this lab, we'll help him figure out as he produces these proteins how those proteins work."
If all goes well, GlycoFi's therapeutic proteins will one day be tested in humans, too. Clinical pharmacologist Lionel Lewis, M.D., is doing the pharmacological and pharmacokinetic analysis of the EPO data. "In the big picture," he says, "I am optimistic that this technology will allow the development of this and many other human proteins that could be synthesized at GlycoFi and then be tested [in humans] for the first time at Dartmouth."
But Gerngross and Hutchinson, for all their enthusiasm, realize there are hurdles to leap before yeast-produced drugs get to the marketplace. "Like eight years of human trials," sighs Hutchinson. But considering that purified insulin had been around for 60 years before scientists figured out how to make genetically engineered insulin, that makes an eight-year timeline seem happily short . . . and sweet.
Carter is the associate editor of Dartmouth Medicine magazine. The subject of this article was unusually close at hand, since GlycoFi is located in the same building as the magazine's offices, so she was able to get a really close-up look at its operations.
If you would like to offer any feedback about this article, we would welcome getting your comments at DartMed@Dartmouth.edu.