MIT researchers meld biology and engineering in the fight against cancer.
Don’t ask Sangeeta Bhatia, SM ‘93, PhD ‘97, whether she’s a biologist or an engineer. For her, the question is beside the point, as it is for all the other MIT professors who use the tools of both engineering and biology to study cancer. “When one focuses on human disease, there isn’t a real boundary between science and engineering and medicine,” says Bhatia, an associate professor who is developing nanoparticles for monitoring and treating cancer.
Bhatia is one of many MIT researchers who will benefit from the establishment of the new David H. Koch Institute for Integrative Cancer Research. This fall, David H. Koch ‘62, SM ‘63, gave MIT $100 million to fund research that brings together biological and engineering approaches to fighting cancer. The Koch Institute will build on the pioneering work of MIT’s Center for Cancer Research; the center’s faculty members (along with 10 MIT engineering professors so far) have become its founding members. Koch’s gift will also speed the construction of a new cancer research facility, which breaks ground this March and is scheduled to open in 2010.
“Bringing in folks from other disciplines is important in the fight against cancer,” says Koch Institute director and biology professor Tyler Jacks, who directed the Center for Cancer Research. “We have the best engineers in the world, so our work force is unmatched.”
Jacks says that viewing cancer as an engineering problem–a relatively new idea–promotes new approaches to basic research and new ways of tackling practical problems. On the basic-research front, it means applying the mind-set and tools of engineering: looking at cells as complex systems and harnessing computer models that can make sense of large amounts of data. On the applied side, it means developing new drugs, new materials for delivering drugs, and new devices for monitoring disease progression.
Many MIT cancer researchers had already adopted this approach in the last five years or so, says Jacks, and the influx of funding and institutional support will ensure that their projects flourish.
“There’s an old-school artificiality to the idea of ‘engineers’ and ‘biologists,’ ” says Dane Wittrup, a professor of chemical engineering and bioengineering and a member of the Koch Institute. To accomplish anything in the biomedical field, he says, engineers “have to have biology in their labs and in their brains.”
Bhatia, who holds an MD in addition to her MIT degrees, agrees. Biologists and engineers have complementary skills; ideas flow in both directions, which is what makes working in both fields so exciting, she says: “We see many often unexpected examples of engineering enabling science through tool development, and science being translated through engineering into diagnostics and therapeutics.”
Here are three projects at MIT that illustrate how melding engineering and biology could help researchers understand–and ultimately cure–cancer.
Dane Wittrup is a protein engineer. He takes cancer therapies and tries to make them work better, with fewer side effects, through a combination of luck and design.
Wittrup makes high-performance antibody therapies that mobilize the immune system against cancer cells, which thrive because the body fails to recognize them as malignant. Classic chemotherapy drugs are “poisons” with harsh side effects, he says. An antibody protein, on the other hand, generally has fewer side effects because it binds to one particular target, such as a tumor antigen. “The immune system sees that [a cell] has antibodies all over it and responds: ‘It’s coated with antibodies, so I’m gonna kill it,’ ” he says.
In order to make better-performing protein therapies, Wittrup uses what’s called directed evolution. “You’re bottling Darwinian evolution and setting the survival rules yourself,” he says. First, he creates tens of millions of mutated versions of the gene for a particular protein and inserts each gene into a yeast cell. Then he uses various screening techniques to identify the yeast cells carrying the best version of the protein. This might be the protein that binds most strongly to a given target or the protein that’s most stable in a particular range of temperatures, among many other possibilities. Wittrup says that protein engineers have refined directed evolution to the point that they can use it to make whatever proteins they want. Now, instead of just working on new tools for designing proteins, he uses such tools to make proteins that could help treat cancer.
“Designing proteins is not the hard part anymore,” he says. “What’s harder and more interesting is determining which proteins to work on, what they do, and how.”
Simply improving one of a protein’s important properties is often not enough to improve its performance as a drug. “Most people are pretty naïve about what’s necessary to make a drug potent,” says Wittrup. The naïve approach, of which he says he has been guilty, is to test therapeutic proteins only for how tightly they bind to their cancer target; but such tests, he says, presuppose that “affinity equals potency.” They are conducted completely out of context, in a drop of cell culture solution in a plastic plate–outside the body, outside the tumors the cells came from.
Wittrup learned his lesson the hard way. He used directed evolution in an effort to improve the performance of an antibody against carcinoembryonic antigen (CEA), a protein that is overabundant on the surfaces of some tumors. The CEA-binding antibody he started out with falls off the cell and is filtered out of the blood before it can prompt a strong immune response. So he developed a version that bound much more strongly to proteins on the surfaces of cancer cells. The binding strength of his engineered antibody was two orders of magnitude greater than that of any other antibody known at the time; by all accounts, it appeared to be a great success. But when he tested it in living mice carrying human colon cancer, it didn’t work any better than the original. How could this be?
Turning back to the tools of engineering, Wittrup and his grad students adapted mathematical models normally used for studying industrial catalysts in order to examine the kinetics of his CEA-binding protein in tumors. In living mice, the engineered antibody was still being disposed of too fast–in this case because it got recycled before it had time to reach the center of a tumor. Proteins cycle rather quickly through all cells, normal and cancerous, generally staying intact only for minutes or hours. They get chopped up into smaller pieces so that they can be reused in new proteins. It’s part of normal cell maintenance; just as your home would be a dump if you never took out the trash, cells would be overwhelmed if they held onto every molecule they ever made or took in.
“Even if the bond lasts a week, it doesn’t matter,” Wittrup says. “It all goes down the garbage disposal.”
Using his models and working with imaging specialists to watch the progress of his protein through live mice, Wittrup is now attempting to modify the protein so that it will diffuse through the tumor too quickly for tumor cells to dispose of it in time. But the CEA-targeted antibody is just one of many therapies he’s developing. Another is a safer version of a toxic drug for kidney cancer and melanoma, which he’s working on with a fellow Koch Institute professor, immunologist Jianzhu Chen.
“We don’t know everything about cancer,” says Wittrup. But he adds that we “know enough to do more” to develop new therapies. Spoken like an engineer.
Cells as Systems
“People say cancer arises from some kind of disregulation of cell function,” says Douglas Lauffenburger, director of MIT’s Biological Engineering Division and an affiliated Koch Institute faculty member. So he’s taking an engineering approach to the question of how cell functions are regulated. His answers could help researchers predict whether a given drug will work.
The Human Genome Project has led biologists to the unfortunate conclusion that there is no simple explanation for how genes interact and how these interactions cause cells to move, grow, and die. Instead, Lauffenburger says, it “alerted everybody to how complex this was going to be.” Researchers had hoped to tackle diseases, including cancer, by tracing them to individual proteins or genes that they could fix with a “magic bullet.” But genome analysis suggests that cell functions are regulated by “lots of genes, lots of proteins interacting.”
Not only does each cell function seem to be affected by many genes, but researchers now think any given function can be the result of many different molecular pathways. Just as some airplanes have more engines than they need, cells have redundancy built in. If a drug interferes with one protein that helps a cancer cell divide, the cell might have recourse to four or five more pathways.
The complexity of these regulatory networks makes it very hard to predict what a drug is going to do, says Lauffenburger. As a result, drug development is “all trial and error”–and extremely inefficient and costly. Researchers test countless compounds to zero in on those with promise. The earlier an ineffective compound can be taken out of the running, the lower the costs of drug development. So in an approach known as systems or network biology, Lauffenburger and others are looking at cells as complex systems and building computer programs that can process large amounts of data about biomolecular interactions. Lauffenburger maps how proteins interact and shows how these interactions influence cell functions, including growth. With such models, he and others hope to predict how drugs will affect cells biochemically and, in turn, how they will affect cell function.
Lauffenburger recently used this approach to predict the effect of a particular drug on epithelial cancer cells, a class of cells that includes those involved in cancers of the colon, breast, cervix, and skin. The drug inhibits one pathway that prevents cell death. This pathway is active in almost all epithelial cancers, so a researcher considering only one pathway at a time would expect the drug to kill multiple kinds of cancer cells. But it turns out that it doesn’t. While the drug increases the death rate of colon cancer cells, it doesn’t increase those of breast or cervical cancer cells, which have several other pathways to fall back on. Lauffenburger’s maps of protein interactions “ascertained how the sum of five pathways works together to govern cell death,” he says.
It’s unclear, however, how broadly applicable Lauffenburger’s models will be. He ticks off the many questions that remain: Will a model of protein interactions in an epithelial cancer cell need many modifications before it can be applied to other kinds of cancer cells? What are the best kinds of measurements to feed into these models? And most important, can the models help researchers uncover why a drug works for some patients and not others? He’s currently pursuing collaborations with the drug companies Pfizer, AstraZeneca, and Merrimack Pharmaceuticals to develop models for use in testing new drugs.
More broadly, the work of Lauffenburger and other network biologists is changing the way biologists look at cancer. “There’s no magic molecule,” Lauffenburger says–no simple, single target to identify that will be the key to treating the disease. But looking at cancer cells as complex systems gone awry, and focusing on what he calls the “actual activities” of their proteins, is leading to a new understanding of how the cells operate.
Today, the only way for doctors to verify that cancer drugs are reaching their targets is to perform magnetic resonance imaging (MRI) scans after weeks of treatment to see if patients’ tumors have shrunk. Sangeeta Bhatia, an associate professor of electrical engineering and computer science in the Division of Health Sciences and Technology and a member of the Koch Institute, is developing multipurpose nanoparticles that she hopes will shorten this process, reduce the side effects of chemotherapy, and make treatment more effective.
Bhatia’s compounds act as precise drug-delivery vehicles and MRI contrast agents; they zero in on tumor blood vessels and, once there, attract more nanoparticles. Recently, she developed a way to make them release their payloads on command when heated by low-frequency electromagnetic waves applied from outside the body.
The nanoparticles are iron oxide spheres bound to tumor-targeting peptides and strands of DNA. The DNA is in turn bound to drugs such as cisplatin, a chemotherapy agent. When the iron oxide cores are heated up by the radio frequency waves (which don’t affect the body’s tissues), the DNA “melts”: the strands of the double helix separate, freeing the drug.
The temperature at which a strand of DNA melts depends on the strand’s length, making the nanoparticles even more versatile. Doctors could administer a cocktail of particles designed to release their drugs at different temperatures, then sequentially activate multiple doses by applying different radio frequencies. What’s more, “the patient wouldn’t have to come in for imaging, then chemotherapy, then repeated scans,” says Bhatia. Because the iron oxide doubles as a contrast agent, MRI scans administered at the time of treatment would be able to verify that a drug had reached the tumor it was targeting.
Other groups are also working on using nanoparticles for targeted chemotherapy. But either their techniques are passive–drug release cannot be controlled but happens over time–or they haven’t worked very well. Bhatia, however, demonstrated her remote-controlled nanoparticles in mice that had model tumors made of gel implanted in them. Radio frequencies applied from outside the mice triggered the release of model drugs that penetrated surrounding tissue.
Bhatia has been working on these nanoparticles for years, in collaboration with Erkki Ruoslahti, a biology professor at the University of California, Santa Barbara, who is an expert on the tumor environment, and with Michael Sailor, a chemistry professor at the University of California, San Diego.
Now, working with MIT Institute Professor Phillip Sharp, she hopes to use the particles to deliver RNA interference therapies, which show great promise but have yet to live up to their potential. In RNA interference, specially designed sequences of RNA hinder the expression of particular genes. Companies such as Alnylam hope to harness the process to therapeutically shut down disease genes, but delivery remains “a critically important problem to solve,” says Sharp, a Nobel Prize-winning biologist and a cofounder of Alnylam. In the technique Bhatia is working on, the nanoparticles would be bound to strands of RNA instead of DNA; rather than carrying a drug, she explains, the RNA would itself be the drug.
Bhatia says the great benefit of her nanoparticles is that they could “integrate detection and therapy.” Doctors treating diabetes and heart disease can already implant glucose pumps and defibrillators that don’t just administer treatment but monitor results, adjusting treatment as necessary. Similarly, Bhatia hopes, the nanoparticles will allow doctors to more quickly assess whether chemotherapy is working and modify it if it isn’t–relieving stress and saving lives.
Become an MIT Technology Review Insider for in-depth analysis and unparalleled perspective.Subscribe today