Lead Story

Breast Cancer Genes:

When the Sequence Is Not Enough

Few cancer genes are more notorious than the genes that cause familial breast cancer—BRCA1 and BRCA2. The New York Times described the cloning of BRCA1 in 1994 as "a genetic trophy so ferociously coveted and loudly heralded that it had taken on a near-mythic aura," but cautioned that since the gene was unexpectedly large, it might take at least a year before a diagnostic test could be developed from it. Fifteen years later, there are indeed genetic tests to evaluate the risk of breast and ovarian cancer in women who possess one of several known mutations. There are, however, even more variants for which the risks are not yet understood. Shyam Sharan, Ph.D., Senior Investigator and Head of the Genetics of Cancer Susceptibility Section in CCR’s Mouse Genetics Cancer Program, understands the difficulties of studying these genes better than many. As a Postdoctoral Fellow, he got caught in the race to understand the BRCA genes by cloning their mouse homologues. That initial sprint turned into a marathon, and although it is far from over, the recently tenured Sharan appears exhilarated by the milestones he has recently passed.

Photo shows Shyam Sharan, Ph.D.
Shyam Sharan, Ph.D. (Photo: R. Baer)

In the four-year period after Mary Claire King’s groundbreaking identification of a region of human chromosome 17 linked to familial breast cancer, BRCA1 and BRCA2 were cloned in humans and mice, and their function was linked to DNA repair. This pace was a source of optimism for the field, intense competitive pressure for the scientists involved, and occasional humor. "I went to a Keystone meeting on breast cancer and gave a talk," remembered Sharan. "After me, Thomas Ludwig also gave a talk about a BRCA2 knockout mouse, and as a joke, I got an award for winning the BRCA2 race by 15 minutes." However, identifying the genes turned out to be only a first step in both understanding their role in tumorigenesis and predicting which mutations would be oncogenic.

In particular, two puzzles from that time have continued to drive Sharan’s research. The first led from the observation that the known mutations did not seem to cluster into any "hot spots" but were distributed throughout the gene, suggesting that all regions of the protein were equally important for tumor suppression. With 1,863 amino acids and 3,148 amino acids respectively, BRCA1 and BRCA2 are huge proteins (for comparison, hemoglobin, which carries oxygen in the blood, has 574 amino acids). Sharan and others had identified associations of these proteins with DNA repair, but that was far from a complete functional explanation of these complex proteins. What did the rest of these proteins do, and how did far-flung mutations contribute to tumorigenesis? The second puzzle stemmed from the seeming paradox that eliminating either protein from mammary cells resulted in cancerous proliferation, whereas disrupting them in embryonic mouse cells resulted in a failure to proliferate and develop. How could a gene involved in something as basic to the cell as DNA integrity cause opposite effects in different cell types?

Human BRCA1 is fully functional in mice, and its expression mirrors the mouse Brca1 gene. In panel a, the Brca1 mutant mouse (right) rescued by the human BRCA1 BAC transgene appears indistinguishable from its wild type littermate (left). Panels b and c show an expression analysis of the human BRCA1 transgene (panel b) and endogenous Brca1 (panel c) in the brain of a 13.5-day mouse embryo. High level of expression was observed in the neuroepithelium (ne) of the ventricular layer (vl) and the external germinal layer (egl) of the cerebellum (cb), as shown by the arrows. Human BRCA1 is fully functional in mice, and its expression mirrors the mouse Brca1 gene. In panel a, the Brca1 mutant mouse (right) rescued by the human BRCA1 BAC transgene appears indistinguishable from its wild type littermate (left). Panels b and c show an expression analysis of the human BRCA1 transgene (panel b) and endogenous Brca1 (panel c) in the brain of a 13.5-day mouse embryo. High level of expression was observed in the neuroepithelium (ne) of the ventricular layer (vl) and the external germinal layer (egl) of the cerebellum (cb), as shown by the arrows.
Image shows human BRCA1, which is fully functional in mice, and its expression mirrors the mouse Brca1 gene. In panel a, the Brca1 mutant mouse (right) rescued by the human BRCA1 BAC transgene appears indistinguishable from its wild type littermate (left). Panels b and c show an expression analysis of the human BRCA1 transgene (panel b) and endogenous Brca1 (panel c) in the brain of a 13.5-day mouse embryo. High level of expression was observed in the neuroepithelium (ne) of the ventricular layer (vl) and the external germinal layer (egl) of the cerebellum (cb), as shown by the arrows. Human BRCA1 is fully functional in mice, and its expression mirrors the mouse Brca1 gene. In panel a, the Brca1 mutant mouse (right) rescued by the human BRCA1 BAC transgene appears indistinguishable from its wild type littermate (left). Panels b and c show an expression analysis of the human BRCA1 transgene (panel b) and endogenous Brca1 (panel c) in the brain of a 13.5-day mouse embryo. High level of expression was observed in the neuroepithelium (ne) of the ventricular layer (vl) and the external germinal layer (egl) of the cerebellum (cb), as shown by the arrows. (Image: S. Sharan, CCR)

Sense through Missense

When Sharan came to NCI, he wanted to use mouse genetics to study the functions of the BRCA genes. He knew that most of the identified mutations in BRCA genes came from tumor samples, and it was therefore not surprising that they resulted in tumorigenesis, but Sharan wanted to be able to mutate regions of interest in these genes systematically to study their effects. However, there was a small problem—the mouse and human BRCA1 genes are only about 60 percent homologous, which means that, in mice, the human gene of interest is already mutated by 40 percent. Nevertheless, Sharan decided to introduce the human BRCA1 gene into mice. And not just the gene, but the entire 200,000 basepair length of human DNA that comprised all of the regulatory elements as well as the gene itself.

"It was kind of risky," commented Sharan, noting that for the experiment to succeed, the mouse cells would need to contain the necessary cellular machinery to properly regulate the human elements, which was by no means clear. Indeed, a paper that came out just as they were making the first mice examined the regulatory elements in a 2,000 base-pair region of the mouse and human genes without finding any obvious conservation between the species. "But we wanted to express the gene at physiological levels and not hook it to a promoter that would overexpress it...and it actually paid off." The human DNA was able to completely mimic—or rescue—the missing mouse BRCA1. Most exciting, the expression pattern of the human gene in these mice was exactly the same as the normal mouse gene, which is expressed ubiquitously in early development and then downregulated in cells that begin to differentiate.

What did the rest of these protiens do, and how did far-flung mutations contribute to tumorigenesis?

However, the goal was to study mutations introduced into the BRCA1 gene. With mouse model in hand, the investigators’ next hurdle to overcome was to be able to make targeted point mutations in a large genetic sequence before creating the mouse. Here, Sharan had the help of his colleagues down the hall—Neal Copeland, Ph.D., Nancy Jenkins, Ph.D., and Don Court, Ph.D.—who had recently developed just the recombineering technology he needed to adapt into his own system (see "Science in Singapore: Aiming High for Biomedical Research," page 26).

"As we started to make mutations, we quickly learned two lessons," explained Sharan. The first was that mutations that were supposed to be deleterious based on their location in highly conserved (and hence arguably important from an evolutionary standpoint) regions of the gene often had no effect on the mice. Even biochemical data showing disrupted protein-protein interactions of the mutated BRCA1 could not predict an abnormal phenotype in the mice. The second thing they learned was that several of the deleterious mutations were a result of altered splicing of the gene, effectively knocking it out completely. So, it was impossible to simply look at the amino acid sequence and predict the impact of a single mutation. Every mutation had to be studied individually.

"You can imagine how this could impact my career and my postdoc’s career. Making mice with no phenotypes is not exactly exciting." The researchers tried everything they could think of to show the effects of their mutations—they aged the mice, made fibroblast cultures from them, and studied them biochemically. And yet, they still found that many of their mutations had no obvious phenotype. They needed to find a better way to screen mutations and to know that what they were looking at were important clues to BRCA1 function and not just a difference between mice and men.

Embryonic Stem Cells Tell All

To generate BRCA1 or BRCA2 knockout mice, the first step was to make mouse embryonic stem (ES) cells in which one copy of the gene is disrupted by gene targeting technology. While they were waiting for the mice, Sharan, at that time a Postdoctoral Fellow in the laboratory of Allan Bradley, Ph.D., at the Baylor College of Medicine in Houston, thought a relatively easy and straightforward next step would be to make mouse ES cells with both copies of the gene missing in order to study the resulting defects. The problem was that he just could not get ES cells that were missing both copies of the gene to survive.

"I wasn’t doing anything with these cells," recalled Sharan, "when I one day realized that they could be a powerful system." He recognized that if he made the remaining mouse copy conditional—so that it could be deleted at will—and added in the human gene sequence, the ES cells would only have one human BRCA gene remaining once he deleted the second mouse copy. How the cells behaved with only the human BRCA gene with or without mutations could tell them a great deal about the individual mutations. If the mutation was neutral, then the cells should survive; if it was deleterious, the cells would die. And with any luck, there would be a range of phenotypes, depending on the specific mutations, that would not affect survival per se but would affect DNA repair or other cellular functions when tested.

Sharan and his Postdoctoral Fellow, Sergey Kuznetsov, Ph.D., first tested founder mutations of BRCA2—those highly specific mutations found in families that have remained relatively genetically isolated—that are strongly linked to breast cancer. As expected, the ES cells did not survive. Then they tried mutations or variants that are frequently found in the general population and are, therefore, thought to be neutral. The ES cells appeared normal. Finally, they tried mutations that they thought might be deleterious based on the available literature. As they had found in their mouse models, the majority did not show an effect in their cell-based assay. "We were kind of depressed," Sharan explained candidly. "We thought that the assay might not be sensitive enough."

Go to page:  1 2