Up, Up, and Array

Luckily, Golub smoothed over the conflict and began spending a day a week at the Whitehead Institute. There, he assembled a group of oncologists and computational biologists to apply the new genomics technology to childhood leukemia, bridging the disparate fields of cancer biology and genomics. “It was the first time anyone had gotten systematic about gene expression–based classification of cancer,” says Golub.

Within just 2 years, he and Lander demonstrated that two types of acute leukemia, which clinicians had spent 30 years characterizing, could be classified based exclusively on their gene-expression patterns. “The fact that one could go in, assume nothing, and rediscover these two subclasses was a proof of concept that one could use unbiased genomic approaches to classify cancer,” says Golub.

Five years later, Golub’s side project blossomed into the Cancer Program at the Broad Institute, a collaborative cancer genomics research center that has engaged in some of the field’s most ambitious projects, including The Cancer Genome Atlas.

“I got lots of advice to come back to home base—to study my favorite gene in great detail rather than frittering away my career on a global view of things.”

Here, Golub looks back at his early career and recalls lessons learned from rejection, the lost art of discovering a gene, and the unexpected challenge of bringing genomics to medicine.

Golub Gets Going

Rejecting rejection. Inspired by his childhood pediatrician to go into medicine, Golub completed an undergraduate degree at Carleton College in Minnesota in 1985 and went straight to the University of Chicago’s Pritzker School of Medicine to pursue an MD. “I was tormented with the medical-school-versus-graduate-school dilemma. But my career path was made simple by the fact that I was rejected just about everywhere I applied. I even got rejected at the medical school that I wound up going to, except that I complained and asked them to take a second look. They said, ‘Okay, if you stop complaining.’ So I went in for an interview. In the end, I got Form Letter A, the rejection letter, and Form Letter B, the acceptance letter, within about 4 weeks of each other. I kept those.”

Bridging the gulf. Golub was lured out of the Midwest to Boston, Massachusetts, for his medical residency. “I had ranked Boston Children’s Hospital as number one. At the time, there was a major gulf between laboratory-based science and the clinic in oncology. There were physician-scientists who were a physician on Monday and a scientist on Tuesday, but the notion of translational research hadn’t really evolved. But Children’s had this commitment of trying to bring research and clinical care together.”

Gene-ius discovery. After his residency, Golub obtained a postdoctoral fellowship at Brigham and Women’s Hospital in Boston. One of his first patients became the focus of Golub’s research: a young man with chronic myelomonocytic leukemia (CMML), a bone marrow cancer that invades the blood. “Cytogenetically, he had a chromosome translocation, but the genes involved weren’t known. We cloned the translocation, which now sounds trivial, but was a big deal then. Half of the translocation involved a known gene, but the other half was a new gene that had yet to be discovered, back when there was such a thing. Trainees in the future will never get to do that, because we have the human genome now.”

ALL aboard. Once Golub had identified the translocated oncogene, called TEL, he began searching to see if it was involved in other leukemias. It turned out that TEL, later officially named ETV6, is the most common translocation in childhood acute lymphoblastic leukemia (ALL), and identification of the gene has influenced clinical care of ALL. “It explains about 25 percent of childhood ALL. That’s a routine diagnostic test now. If you have this particular gene translocation, you get less chemotherapy compared to other ALL patients.”

Far and wide. By 1999, Golub was working part-time at the Whitehead Institute, and he and Lander published their seminal paper in Science demonstrating that genomic technologies are a powerful tool in leukemia classification. He didn’t stop there. “We set out in collaborative studies to do gene expression more broadly over the ensuing years, and applied the approach to a number of different tumor types. We discovered expression patterns that had diagnostic potential or offered insights into the pathogenesis of the disease that weren’t otherwise obvious.” Golub was convinced of the potential for genomic technologies to shed light on cancer, but not all his colleagues agreed. “I got lots of advice to come back to home base—to study my favorite gene in great detail rather than frittering away my career on a global view of things.”

Broad intentions. In 2004, Golub joined Lander to found the Broad Institute of Harvard and MIT, a collaborative center designed to tackle big problems in genomics and medicine. “The inspiration was to bring together scientists of different disciplines to take on some audacious goal. In my case, bringing together oncologists and genomicists was really powerful. In order to achieve a goal, you need to tackle something beyond your own know-how.” For example, the Broad has been a leading player in The Cancer Genome Atlas, a National Institutes of Health project to systematically characterize the genetic causes of cancer. “You need more than genome sequencers to do that. You need deep expertise in genome-sequencing technology, computational-analysis methods, cancer biology, and clinical oncology. There’s no individual or group on the planet that has expertise in all those areas. You need a place that can build teams around those kinds of ideas.”

Diversity matters. “I have a very diverse lab: most people in the lab have a different project. Increasingly, we’re orienting the lab toward therapeutic potential, which is the way the field is moving right now. We’re trying to develop innovative methods for drug discovery. With Kim Stegmaier, who was a postdoc in my lab and now has her own lab at Dana-Farber, we’ve shown that one can use this gene-expression signature approach to discover chemical compounds that induce the differentiation of cancer cells. It turns out that many types of cancers are stuck in an immature state, and if you could figure out how to coax them into maturity, maybe they would kill themselves like normal cells. So we screen for chemicals that cause that differentiation signature to turn on, and those are candidates for clinical trials.”

Golub Gets Philosophical

Science sleepovers. “As a college student, I had no idea what serious biomedical research was supposed to be like. Through a family connection, I found myself in a small, hospital-associated lab where I was given lots of room to run with a project. I was trying to make in vitro models of cervical cancer development, before it was clear that HPV causes cervical cancer. That’s where I found the idea that observing something in nature for the first time was pretty exhilarating. I remember setting up a cot in the lab so I could stay there all night to do time-lapse photography manually.”

Fresh slate. That early, independent lab work left a lasting impression on Golub. “That freedom to go off and discover, without being encumbered by the fact that I actually didn’t know very much, was a pretty influential experience. I’ve come to believe that having a deep fund of scientific knowledge isn’t as important as many might believe. I don’t think it’s a prerequisite for creativity in science. A lot of really important work comes from creative new approaches that don’t have a prerequisite of encyclopedic knowledge of what’s come before. In fact, that stuff can sometimes get in the way. It constrains your thinking as much as it supports a new idea.”

Paper value. In 1994, Golub sent the paper on his postdoctoral research identifying the gene responsible for CMML to Science. “They sent it back without review, declaring it totally uninteresting. We took the same paper and sent it to Cell. The response there was that they loved it and wanted to have it. That was a good life lesson in scientific publishing. I tell trainees in my lab that they can’t take this stuff too seriously. How your paper does is not an intrinsic measure of your value and worth as a scientist.”

Left behind. “The combination chemotherapy approach was first worked out in childhood leukemia. Pediatric oncology was a real leader in demonstrating this concept of combination therapy to avoid drug resistance, and today 80 percent of [pediatric] patients can be cured with these collections of old-fashioned drugs. Some would view that as fantastic—80 percent success—and that is obviously good, but it is still 2 years of awful therapy, late side effects such as learning problems, and social and developmental consequences of taking 2 years out of a family’s life. So I reject the notion that childhood leukemia’s problem has been solved and we should move on to the next big thing. I think childhood cancer is at risk of moving from a leader to a laggard. The rest of the field is moving on to treating subsets of lung cancer with [epidermal growth factor receptor] inhibitors instead of chemo, and we’re not on pace to do that with most childhood cancers.” (See “After Chemo” for more about late side effects of chemotherapy.)

Slippery signatures. “It turned out the translation of gene expression signatures into commercial diagnostic tests for cancer is technically difficult. It’s fair to say in general that there’s only a small number that have made their way to routine practice. There’s this paradox that while cancer is a big public-health problem, enough samples to fully validate a specific test are hard to come by. In addition, one of the challenges of gene expression is that you’re looking at the average expression across a tissue sample. In real life, biopsies are a mixture of tumor cells and normal cells, and that makes the signatures complicated.”

On the hunt. “There is real power in starting from patients and working back to the genetic causes of a disease, as opposed to starting with a hypothesis and going in search of supportive evidence. It’s not a gene in search of a disease; it’s a disease in search of a gene. That’s very powerful. You don’t know what you’re going to find, but whatever it is, it’s going to be important.”

Golub Gets Refreshed

Setting goals. “I value bringing together a community of researchers and creating a culture of collaboration that can have more of an impact on our field than the accomplishment of an individual. I hope there is now a sustaining culture of scientists helping each other and keeping their eye on changing the world. That’s the goal. Being first author on a paper is not the goal. The goal is to change the world.”

“There is real power in starting from patients and working
back to the genetic causes
of a disease.”

Artist-in-residence. “When the Broad was first starting, we had some space and invited in a painter, Daniel Kohn. We had a studio for him in the lab, so he was painting next to people doing experiments. We’ve since had two additional artists-in-residence. Great science in my view is about creativity, and if that’s the case, surrounding yourself with other creative people is a way of stimulating your thinking, to think about old problems in new ways. What we do in science isn’t tackling new problems, but looking at old problems with a new lens. That’s what artists do—seek new lenses to see the world through. Some scientists think it’s totally crazy. They ask, ‘Why is there a painter in our midst?’ But I don’t mind being a little provocative.”

Seeking simplicity. “My favorite place outside Boston is the Berkshires in Western Massachusetts, where we have a little one-room cabin that we built a few years ago. When I get on the Mass Pike heading toward this cabin, I feel the stress of grants, work, and my blood pressure recede. I don’t get there as much as I’d like to.” 

Greatest Hits

• As a postdoctoral fellow, identified a gene responsible for 25 percent of childhood cases of acute lymphoblastic leukemia

• Working with genomicist Eric Lander, laid the foundation for the diagnosis and classification of cancer using gene-expression profiles

• Developed a high-throughput method to screen for cancer drugs based on gene-expression signatures

• Established an ongoing artist-in-residence program at the Broad Institute to inspire creativity and crosstalk between science and the arts