Photo by Todd McNaught
A detective has a slim chance of catching a crook who escapes without a trace. But an investigator who can track a culprit's friends and associates might learn a lot about the criminal's next target.
This "guilt-by-association" strategy turns out be just as effective in cancer research as it is in law enforcement, since cancer-causing proteins that rob cells of their ability to divide normally also depend on co-conspirators. Such molecular partners in crime, once identified, may yield a wealth of information about how a cancer develops and how it best can be treated.
Research fellow Dr. Harry Hwang, working with Dr. Bruce Clurman of the Clinical Research and Human Biology divisions, and colleagues, successfully used this approach to identify cancer genes that collaborate with a protein called p27, which shows up in abnormally low levels in lymphomas and breast and other cancers.
21 potential cancer genes
The team identified 21 potential cancer genes, 14 of which had not previously been implicated in the process. Made possible by the recent completion of the mouse and human genomes, the achievement represents a key step toward the quest by cancer biologists to characterize the cellular pathways that distinguish one type of cancer from another.
The study was published July 31 in the online early edition of the Proceedings of the National Academy of Sciences. Co-authors include Dr. Matthew Fero in the Clinical Research Division and Dr. Anton Berns and his colleagues at the Netherlands Cancer Institute in Amsterdam. The study was funded by the National Institutes of Health and the W.M. Keck Foundation.
"We know development of cancer requires multiple steps," said Hwang, an E. Donnall Thomas scholar of the Jose Carreras Foundation. "Based on work in some human tumors, it appears that loss of p27 is one significant event. We asked the question, what are the other players in the pathway that lead to tumor formation when p27 is involved?"
Identification of the complete network of genes required for tumor formation, Hwang said, is a first critical step toward developing specific anti-cancer therapies.
The researchers focused on the p27 pathway because the protein has been implicated in the development of many cancers, said Clurman, also an assistant professor of medicine at the University of Washington.
"In normal cells," he said, "the p27 protein blocks cell division in response to many signals, and p27 functions as a tumor suppressor. That is, the loss of p27 function is associated with the development of cancer. But how p27 works to suppress cancer is unknown."
Because normal cell division and its perturbation in cancer are complex processes, the team reasoned that much could be learned from identifying proteins in the cell that collaborate with p27 in tumor development.
Random insertion of DNA
To identify potential new cancer genes, the team took advantage of a virus known as the Moloney leukemia virus, which causes lymphomas in mice. Upon infection, the virus inserts its own DNA randomly into the DNA of the cell. As a result, host cell genes adjacent to the site of viral insertion become activated. In rare cases, the gene near the viral insertion may be what is known as a proto-oncogene, a gene that can promote tumor formation when activated.
The benefit of this technique is that the integrated virus serves as a molecular tag that eases later isolation and characterization of the genes next to the viral insertion.
Erin Randel, former research technician in Fero's lab, infected normal mice and mice deficient in p27 with the virus. She and Fero observed that although tumors form in both groups of mice, the process is greatly accelerated in the p27-deficient animals. Hwang then used genome sequence information to determine the identities of the genes next to sites of viral insertion in these mouse tumors.
Since the virus inserts itself randomly in the mouse genome, the researchers expected to find many potential cancer genes whose activation led to lymphoma in both groups of animals. Of most interest were insertion sites common to more than one p27-deficient mouse. With tens of thousands of genes in the mouse genome, insertions in a common site in multiple p27-deficient animals (with few or no insertions in the same region in normal mice) are likely to be near genes that play the most critical roles in p27-mediated cancer development.
"Insertion sites common to two animals are like pieces of gold," Hwang said.
Among those most exciting identified in the study, he said, are two sites, each of which occurred in seven p27-deficient animals and rarely in the normal mice.
One of this pair is near a gene called Jundp2, which forms a protein known to interact with another cancer-causing gene called Jun. On its own, Jundp2 was not previously implicated in causing cancer. The other insertion is in a region of the X chromosome of the mouse, where genes have not been well characterized.
Figuring out the identity of the genes near insertion sites was greatly facilitated by the recent completion of the mouse-genome sequence, Clurman said.
"The technique of insertional mutagenesis with the Moloney virus, combined with the resources of human and mouse genome sequences, allows one to start with one defined mutation - in our case, p27 - and identify an enormous number of genes involved in cancer," he said.
"The old approach to doing this was a tremendous fishing expedition. With the genome sequence, you have the direct address of a targeted region, which allows us to discover cancer genes involved in multiple tumors that we would have never found using old methods. This genomic approach has revealed that a lot more genes are involved in these lymphomas, and probably in other cancers induced by these types of viruses, than had been known previously."
Clurman's group next plans to verify whether the genes they've identified are indeed activated in their set of experimentally induced mouse lymphomas. If so, based on the similarity between the human and mouse genome, they can begin to examine whether human tumors contain hyperactivated forms of these genes as well. One or more of such cancer genes in a given pathway might serve as suitable targets for new cancer therapies.
"We might not be able to develop a therapy that targets p27 directly," he said. "But we may be able to target one of the collaborating proteins or pathways."
The real power of this approach, Clurman said, will be when he and others in the field amass a large collection of the components of many different cancer pathways.
"Other labs are looking for collaborators of genes associated with different cancer pathways," he said. "A huge database will be built based on this information, which will let us compare the sets of genes altered in different kinds of tumors. Gene discovery is just our first step toward developing new approaches to treat human disease."