In the late 1990s, Todd Golub developed a technique that has revolutionized cancer diagnosis and treatment. He showed that DNA chips, which measure the activity or expression of thousands of genes simultaneously, could discriminate between two similar leukemias because each cancer has a unique genetic fingerprint.
The gene profile approach has proved so significant that doctors and physicians are currently using multigenic or genomic approaches to differentiate associated cancers and to determine the prognosis of a particular cancer. Validation of gene expression signature techniques in patients is ongoing.
Golub, now the director of cancer genomics at the Broad Institute, which he helped to establish in 2003, continues to advance genomic methods to improve the understanding, diagnosis, and treatment of cancer and other diseases. He also is actively involved in moving genomic technology from the research laboratory into the clinic. Broad, a collaboration of the Massachusetts Institute of Technology and Harvard and its affiliated hospitals, aims to use genomics to transform medicine.
Genomic approaches, Golub says, can probe human disease in ways never before possible in medicine's history. "You can only be so smart in generating a hypothesis based on understanding a biological process and using experimental model systems," Golub explains. "It can be equally or perhaps more powerful to read from the human genome what the abnormalities are in a given tumor or disease state."
The ability to tailor therapies to individuals based on their unique genetic endowments should also result from genomics, he says. "We need to move toward personalized medicine," Golub says. "In cancer, that means giving the right drug for a given cancer rather than a generic chemotherapy."
Golub, who graduated in1989 from the University of Chicago Pritzker School of Medicine, first identified a genetic cause for a cancer, acute lymphoblastic leukemia (ALL), as a Dana-Farber Cancer Institute research fellow in the early 1990s.
During the then pregenomic era, when technology limited analysis to one gene at a time, Golub identified the TEL gene, an aberrant fusion of two genes from a chromosomal abnormality, as causing ALL in one of his patients. TEL identification now informs treatment of ALL: the 25 percent of patients with the mutation need limited chemotherapy compared to other ALL patients.
When DNA chips became available in the mid 1990s, Golub did his pioneering work distinguishing leukemias. Since then, he has classified many tumors, including brain, lung, and prostate cancer, by their genetic profiles. He also has found genetic signatures of metastasis, when tumor cells fatally spread in the body. Recently, he revealed how problems with noncoding RNAs, RNAs that do not make proteins, play a role in cancer.
Besides working on improving cancer diagnoses, Golub employs genomics to screen for new drugs and to investigate biological processes causing disease. To screen drugs, first Golub makes expression profiles of 100 selected genes for a diseased cell growing in culture and for its healthy counterpart. He then treats a sample of the diseased cell's different molecules—one at a time—and assesses each of the resulting expression profiles. Compounds that make the gene pattern resemble the healthy one become candidates as possible drugs, he says. The process, which is automated, is called gene expression–based high-throughput screening, or GE-HTS.
"If GE-HTS works, it will be a fundamentally different way of approaching drug discovery," Golub says. With GE-HTS, Golub found an agent that may halt a leukemia cell from growing uncontrollably. He also is using GE-HTS to study how lithium works in bipolar disorder.
In 2006, he extended GE-HTS by developing expression patterns of all the genes in the genome in a handful of cells in response to 1,500 drugs. He then put the 1,500 patterns in a searchable database called the Connectivity Map.
Using the map, a scientist can now treat cells with a drug with an unknown mechanism of action, obtain the drug's gene expression profile, and then compare its profile with those in the database, Golub says. "The software ranks results based on similarities to the profile given to it," Golub explains. The subset of genes found could reveal information about the potential therapeutic application of the drug.
Similarly, a disease state has a gene expression profile, Golub adds. "You can then ask the Connectivity Map to show you a drug with a profile that, perhaps, is the reverse of the diseased profile," he says. That drug might reveal a possible treatment target for that disease.
Golub credits a willingness to embrace creativity as the foundation of his science. Although many people, he says, think science requires rote learning of facts and proscribed methodologies, Golub says he tries not to let convention hamper him too much. He even created an artist-in-residence program at Broad to help scientists and artists question assumptions about their work.
Golub believes the process of creating art and science is similar. Both try to describe reality and are limited in their ability to do so: "In the visual arts, say, by canvas size," he says. "In science, by technology or resources." Golub, however, keeps pushing to find ways to overcome constraints in science's ability to render the complexities of the human genomic landscape.