Gustavo Leone, PhD, OSUCCC – James associate director of Basic Research
When a family of transcription factors failed to perform as expected in animal experiments, an Ohio State cancer researcher began a series of painstaking genetic studies to learn what was going on.
By Darrell E. Ward
Photography by Roman Sapecki
Note: Human genes are spelled in uppercase italics. animal genes are in mixed case. Gene products are in normal type.
When researchers at the Massachusetts Institute of Technology isolated and cloned the gene responsible for retinoblastoma in 1986, it was a stunning breakthrough.That gene, RB1, was the first example of a tumor-suppressor gene, "that priceless category of genes that, among other tasks, protect us from developing cancers," as Nicholas Dyson described them in a 2003 commentary in the journal Nature.
But how did this tumor suppressor gene work and how did its loss contribute to cancer? Many thought that the answers to those questions were well in hand following 10 years of mainly cell culture in vitro experiments. Then Ohio State cancer researcher Gustavo Leone—who had made significant contributions to those cell-culture studies—began verifying the findings in animal models.
"That's when things got interesting," says Leone, PhD, associate director of Basic Research at the cancer center.
Getting it right was important. "Understanding RB and its fellow molecules is critical for understanding cancer," says Michael Ostrowski, PhD, chair of Molecular and Cellular Biochemistry, co-leader of the Molecular Biology and Cancer Genetics research program, and one of Leone's collaborators. "Mutations in both copies of RB1 lead to retinoblastoma, a malignancy in children, and families with RB1 mutations are predisposed to cancer. The RB pathway is also inactivated in most human cancers, so knowledge of the molecular interactions involved is central to cancer biology."
Basic research often begins with little knowledge of the thickets that lie ahead. The cloning of RB was followed a few years later by the discovery of a protein that binds both the RB1 protein and the early 2 (E2) protein encoded by adenoviruses. The new molecule was called E2 Factor 1, or E2F1. It was a transcription factor, a protein that binds with specific DNA sequences to activate or repress genes. A Duke University group found the protein; a Harvard group cloned the gene.
ALL IN THE FAMILY
The discovery of E2F1 soon led to the discovery of a family of genes: E2F2 through E2F6. Cell culture studies revealed that the E2F1, 2 and 3 proteins were gene activators and the only ones that bind with RB1 to regulate the cell cycle.
In 1990, Leone, then a graduate student at the University of Calgary, was attending an international virology conference in Berlin where he heard talks by the groups from Duke and Harvard about their discoveries.
"The investigators showed that RB bound to the E2F protein, and that the two of them together somehow executed some function in cancer," Leone says. "They also had found that three cancer-causing viruses—adenovirus, papillomavirus and polyomavirus—encoded proteins that could bind RB1 and release an activity that people didn't understand."
Leone was fascinated. "I thought, 'Wow, I'd like to work on that.'" He joined the Duke group as a postdoctoral candidate in 1994 and has studied the E2F family ever since.
At Duke, Leone studied the similarities and differences among the family members, work that helped establish their importance in cell-cycle regulation. "The evidence suggested that E2Fs played a straightforward role in regulating the cell cycle," he says. Gene activators E2Fs 1, 2 and 3 advanced the cycle, and E2Fs 4, 5 and 6 suppressed it. (At Ohio State in 2003 and 2005, his lab discovered E2F7 and E2F8 respectively, the last of the family. They, too, encode repressor proteins.)
The link between E2Fs and cancer is more tenuous, Leone says. "We need to know how these molecules function normally. If we don't know that, it's hard to imagine what they might be doing in cancer."
The mammalian E2F family of transcription factors. The members occupy eight chromosomal locations that encode nine distinct proteins. Traditionally, the family has been divided into gene activators (E2F1 to E2F3) and repressors (E2F4 to E2F8). A few key facts about the family:
• The first six E2Fs encode proteins that bind to DNA only after coupling with a second protein, called a dimerization partner protein (DP).
• Only the proteins encoded by E2F1 to E2F3 bind with the retinoblastoma protein, RB.
• The repressors E2F7 and E2F8 are structurally unique.
• The E2F7a and E2F7b proteins are produced by alternate splicing of the E2F7 primary messenger RNA. The two forms of the E2F3 protein (E2F3a and E2F3b), on the other hand, are transcribed from two distinct promoters within the gene.
CONTRADICTIONS OF INTEREST
When Leone joined The Ohio State University Comprehensive Cancer Center – Arthur G. James Cancer Hospital and Richard J. Solove Research Institute (OSUCCC – James) in 1999, he and his laboratory set out to verify what was known about the family in vivo, in animal models.
"We generated mouse strains that had mutations in each of these genes, and we began looking at their role in normal development and in cancer," Leone says. Based on the earlier in vitro studies, they had an idea of what these experiments should show. "But we observed nothing like what we expected," Leone says. "For example, loss of the genes sometimes had little or no effect on cell growth during development, and in other cases it had a dramatic effect but not the type that we had predicted.
"What we had supposed was true based on cell-culture experiments was not the full story," Leone says. "In fact, some of it was wrong, so we began revisiting the initial questions much more rigorously."
They designed experiments that were as thorough as possible. Focusing only on the gene activators—E2F1, E2F2 and E2F3—Leone and his collaborators generated double- and triple-knockout mice that lacked different combinations of the three genes. In some strains they used conditional gene targeting to knock out E2Fs at will.
Their findings, published in Nature (2001), showed that mice underwent normal embryonic development in the absence of E2f1 and 2, but that E2f3 was required.
Furthermore, cells from the embryos that lacked E2f1 and 2 proliferated robustly in culture, but those without E2f3 did not proliferate. Overall, the study provided the first genetic evidence that these three genes play an essential role in cell-cycle progression, proliferation and development. The work succeeded because the conditional gene knockout strategy enabled the researchers to overcome the lethal effects of inactivation of multiple E2Fs in earlier animal models.
WHY SO MANY?
Nematodes and insects have one E2F activator and one repressor. More complex animals have families of transcription factors. Why do mammals have a large family of E2F genes to accomplish what just one activator and one repressor could do?
The same question applies to other transcription factors, Leone says. "Why are there so many members in the p53 family, the MYC family and the ETS family? The assumption was that each E2F family member regulates specific genes."
The function of E2F family members grew more complicated following a study published in Molecular and Cellular Biochemistry (2000) for which Leone was first author. It showed that E2F3 actually had two distinct promoters and expressed two related proteins, E2F3a and E2F3b.
"We wanted to learn how the three activator genes functioned in concert during development and to dissect out the 'a' and 'b' dichotomy," he explains.
The three genes were clearly working to the same end, but were they doing the same thing? "Each family member can carry a bucket, and together they will fill the tub," he says. "Or each could do a specific job—one holds the bucket, one fills the bucket, one carries the bucket, and one empties the bucket—and work in concert to fill the tub."
The ongoing thinking was that each of the genes did something different. "We used combinatorial knockout models to investigate that question at the genetic level." Their findings, published in Nature (2008), showed that E2f3a is the most important gene. Its presence alone was enough for development through adulthood, Leone says.
But what made 3a so special? "This was the exciting part of the study," he says.
The investigators swapped one family member for another. They took out the E2f3a and replaced it first with E2f3b and then with E2f1. "If 3a was doing something really special, neither 3b nor E2F1 should be able to replace it; the embryos shouldn't develop. Well, in both cases, the animals developed just fine."
It turned out the 3a protein wasn't special at all. "What was critical was the regulation of the 3a gene locus—the timing of its activation and the levels of expression," Leone says.
Still, theE2Fs may not be essential for development, but they might be required for long-term survival. "We think that mammals have so many of these genes because they are needed for living long term," he says. They are investigating that question now.
Tantalizing findings published in Nature (2009) by Leone and his colleagues could have important implications for the role of E2Fs in cancer.
That study shows that E2F 1-3 are, in fact, activators in stem cells, but that their role changes as cells differentiate. In differentiating cells, the E2F activators switch roles and become repressors. "Their function as gene repressors in differentiated cells is opposite what we thought for nearly 20 years," Leone says.
The study also shows how the switch is made. Finally, the researchers show that when RB1 is lost in mature cells—as happens in most cancers—these same E2fs once again become activators.
"In normal cells, these E2fs have a repressor role, but in cancer cells they are activators." There, they could provide a new therapeutic target, he says. "If we can inactivate these E2fs in cancer cells, we may prevent cancer cell proliferation with few major effects on normal differentiated cells."
The findings provide a better understanding of cell proliferation and death. In a developing animal, the change of E2fs from activators to repressors allows stem cells to make the transition to differentiated cells.
"This is important for differentiation," he says. "These E2fs regulate the proliferation that parallels differentiation."
They observed the switch in the differentiating cells of the intestinal crypt, Leone says. "If it doesn't happen as these cells differentiate, they accumulate DNA damage and start dying."
Leone likened the parallel nature of the two events to turning off the lights and closing the door when you leave the room. "To leave the room correctly, you need both
actions, which are independent of one another."
Similarly, differentiating cells need to exit the cell cycle, and these E2fs are important for that. "Before this, these molecules were thought to be important only in proliferating cells like stem cells," Leone says. "Because differentiated cells don't proliferate, these activators were thought to be irrelevant. But we show that that's not the case. It's just the opposite."
While examining embryos during the E2f studies, an observant postdoctoral student in Leone's lab, Alain de Bruin, realized that the placental tissue from Rb-negative mice was highly disorganized compared with wild type placentas. The layer of the placental wall where oxygen and nutrient exchange occurs was severely disrupted.
"The question was," Leone says, "were the problems of the fetuses in the Rb-deficient animals a side effect of placental disruption, or were they due to Rb deficiency? Numerous previous studies attributed the problems to Rb deficiency."
Leone and his collaborators reported their findings in Nature (2003), noting that the layer's trophoblast cells were poorly differentiated, neoplastic and dying by apoptosis. Placentas that were Rb-negative also showed poor transport of essential fatty acids and decreased surface area for oxygen and nutrient exchange.
Next, the researchers used a conditional gene-knockout model to restore Rb function in the placenta and this avoided many of the embryonic abnormalities seen in the Rb-deficient fetuses. The findings showed that the lethal effects on Rb-deficient embryos are due to placental abnormalities and not to the Rb deficiency.
"We showed for the first time that Rb is important in the placenta," Leone says.
In a follow-up study, Leone and his collaborators used a mouse strain that permitted conditional deletion of both Rb and E2f3. That study, published in Genes and Development (2006), showed that during mouse placenta development, Rb has a crucial function in placental stem cells, i.e., trophoblast stem cells.
Furthermore, deleting both the Rb and E2f3 genes enabled the embryos to live three additional days and reduced trophoblast cell proliferation. "There's obviously still something wrong, but it extended gestation," Leone says. "This strongly suggests that E2f3 protein is an important partner in these responses."
The findings have important implications for cancer, which sometimes arises from a select few cells that have characteristics of stem cells, he says. "Here we show in a developmental system that the function of a major tumor suppressor is important in a stem cell compartment, and that the function of the gene is different in stem cells compared with their derivative cells. This adds to the evidence that tumor suppressors are important in stem cells."
The finding in the placenta was completely unexpected. It told Leone and his collaborators that genes can have distant effects. "It told us to think broadly, and that led us to the microenvironment and the idea that genes inside tumor cells can influence surrounding cells in ways that favor tumor growth."
In one of those studies, Ostrowski and Leone showed in Nature (2009) for the first time that gene changes in normal tumor fibroblasts foster tumor growth and progression. The work, which involved loss of a gene called Pten from normal mammary fibroblasts in mouse tumors, also provided the first animal model that accurately represents the microenvironment within human breast tumors.
"We found that normal stromal fibroblasts play an important role in suppressing cancer development and may explain why some human breast cancer patients respond to a standard therapy while others with apparently identical disease do not," says Ostrowski, the co-principal investigator on the study with Leone.
The study also identifies new bio-markers specific to the fibroblasts that may help guide breast-cancer therapy and new molecular targets for developing therapies aimed at gene changes in stromal cells. The findings might also improve the understanding of other pathological conditions influenced by the tissue microenvironment, such as autoimmune disease, lung fibrosis and neurodegenerative diseases.
"Our findings reveal a new role for this gene in the tumor environment, which could lead to entirely new treatments for breast cancer and perhaps other solid tumors using agents that target stromal cells," Leone says.
THE ROAD FROM HERE
Leone's lab is now exploring the many functions and interactions of the E2F activators and repressors using a systems biology approach.
"We are using genomics, protein binding, genome and chip sequencing and gene expression profiles," Leone says. "We'll first use genomics and expression profiling to narrow
down which combinations are probably working together in orchestration and then focus on those."
Their goal is to learn how the entire E2F program is coordinated to regulate cell proliferation in vivo. "We want to identify and understand all the genes in this network, how they function and how they influence the tumor microenvironment," Leone says.
How does the RB tumor-suppressor gene work and how does its loss contribute to cancer? The answers are still not "well in hand," though they are "better in hand." On a broader scale, the effort to understand RB illustrates that cancer yields its secrets only to thorough, sound science.