The unusual suspect

Cancer researchers look beyond the genome to the epigenome—and the role of methyl marks

More than 10,000 people in the U. S. will die of kidney cancer in 2007, the American Cancer Society predicts.

And often, it won’t be genes gone bad that get them.

Scientists have known for decades that cancer can be caused by genetic mutations. But they’ve recently discovered another culprit—the tiny methyl group, which consists of a single carbon atom surrounded by hydrogen atoms. When this chemical group shows up in the wrong places on an otherwise normal strand of DNA, it can cause cancer.

In one breed of mice, intestinal tumors form in two distinct stages, which are partially regulated by epigenetic events, including misplaced methyl groups. First, microscopic tumors, such as the one in the center of the top image, develop. Given the right circumstances, these growths progress into macroscopic tumors (bottom). Photos: Yasuhiro Yamada

In 1994, two groups showed that about 57 percent of patients with the most common form of kidney cancer harbor a mutation on the von Hippel-Lindau (VHL) tumor-suppressor gene. This finding led some doctors to wonder about the remaining 43 percent—how do they arise?

Stephen Baylin, professor of oncology at the Johns Hopkins University, and his colleague James Herman, now an associate professor at the same institution, decided to delve deeper into this medical mystery by taking a closer look at the VHL gene in patients with the non-hereditary form of this cancer.

His lab uncovered an interesting pattern. In roughly 20 percent of the tumors, the DNA bases forming the VHL promoter (the region where proteins bind to activate the gene) have acquired extra methyl groups. However, the sequence of the DNA bases in the whole VHL gene is usually normal, indicating that the gene has not suffered a mutation. Baylin and his colleagues hypothesized that the extra methyl miscreants were guilty of shutting down the otherwise normal gene. Scientists had already shown that methyl groups block access to DNA, preventing it from being read out, so this was a logical conclusion.

Although Baylin was hardly the first scientist to observe odd methylation patterns in the DNA of tumors, he was among the first to produce evidence that this might play a major role in cancer formation. A deluge of papers came out around that time, including a key one by Whitehead Member Rudolf Jaenisch, providing irrefutable proof that misplaced methyl marks can contribute to cancer formation.

“I think the VHL gene was precociously trying to tell us something,” Baylin says. “If you find a gene that has lost its function via a mutation, then you can probably find cases where that gene has lost its function via a modification to the epigenome.”

Marking up DNA—and passing it on

So what’s the epigenome?

You can think of it as the system that lets each type of cell access parts of the genome for its own particular needs.

Strategically placed methyl groups (shown in red) block access to key regions of DNA, keeping specific genes silent. There are two known types of epigenetic marks—methyl groups and DNA-packaging proteins. To see how methyl groups work on one gene, see Whiteboard (0.2 mb pdf or 6.8 mb jpg). Illustration: Christina Ullman

The epigenome serves as a firewall, hiding certain genes while exposing others. For example, a few methyl groups on the promoter of a gene can keep it concealed and silent in a particular tissue. Though methyl marks are the best understood epigenetic marks, there’s another major group—packaging proteins. For example, some proteins block access to genes by coiling bits of the sequence into neat “spools.”

Epigenetic mechanisms usually help cells express genes at the right time and place. While all of an organism’s cells share the same genes, epigenetics ensures that a brain cell produces dopamine, serotonin and other “brain” chemicals rather than keratin, fats and oils, which are characteristic of a skin cell. At least 200 different types of cells comprise a human being, and each one contains a different epigenome.

Given their essential functions, epigenetic marks hardly serve as DNA accessories. But they can be changed like a pair of earrings or a necklace. For example, an enzyme called Dnmt3a places methyl marks on previously unmethylated DNA. Typically active in developing embryos, this enzyme helps to establish tissue-specific DNA methylation patterns.

Importantly, such marks are replicated during cell division and passed to daughter cells. Thus epigenetic marks are transient in one sense, yet heritable in another.

“This dichotomy is one of the reasons we’re studying epigenetics as it relates to cancer,” says Heinz Linhart, an MD/PhD in Jaenisch’s lab. “Epigenetic marks provide potent therapeutic targets because they can be added or stripped, but we wouldn’t be talking about them if they weren’t heritable. Neither mutations nor misplaced methyl marks would induce tumors if they were diluted out when cells divided.”

Linhart manipulates the epigenomes of mice to explore methylation patterns previously linked to tumor formation. He tinkers with methyl marks and watches the results—an approach that allows him to establish cause and effect.

Jaenisch used a similar approach more than 10 years ago to silence critics of the first studies that provided evidence that epigenetic changes can produce tumors.

Tale of a tumor

In 1994, the same year Baylin completed his kidney cancer study, Jaenisch began to study methylation in tumor-prone mice. Though healthy in most respects, these animals develop large numbers of tumors in their intestines.

Ironically, Jaenisch suspected that missing methyl groups might be to blame. Around 1980, scientists had noticed that the DNA of many tumor cells was missing methyl marks, but they didn’t have the tools to probe the relationship. Furthermore, renowned cancer researcher Bert Vogelstein of Johns Hopkins had observed this “hypomethylation” pattern in the tumor-prone mice and proposed that it was a prerequisite for polyp formation.

Eva Moran and Heinz Linhart manipulate mice epigenomes by tinkering with methyl marks, and then see how that affects tumor formation. Photo: Kim Furnald

In collaboration with Whitehead Member Robert Weinberg, Jaenisch and postdoctoral researcher Peter Laird (now a professor at the University of Southern California) stripped methyl groups from the DNA of their pint-sized subjects and waited for the animals to develop tons of tumors.

The results, published in Cell in 1995, were startling.

Rather than mimicking tumor formation, these mice produced fewer tumors. “Though we were puzzled by the outcome, we were pleased to establish a causal relationship between methylation and cancer,” says Jaenisch, who also is a biology professor at Massachusetts Institute of Technology.

A decade later, Japanese pathologist Yasuhiro Yamada joined the lab. Yamada was particularly knowledgeable about the mice used in the experiment. He knew that their intestinal tumors developed in two distinct stages. First come microscopic tumors that resemble flowers. Given the right conditions, these grow into massive irregular tumors that can be seen with the naked eye.

Yamada repeated the 1995 study and discovered that hypomethylation increases the number of tiny tumors but decreases the number of large tumors. The earlier researchers missed the microscopic effect.

“Our lab had just shown that global hypomethylation destabilizes DNA big time, so we reasoned that the small tumors result from chromosomal instability rather than epigenetic silencing,” Jaenisch explains.

Linhart and MIT diploma student Eva Moran took the study one step further by setting new methyl marks randomly on the DNA of the tumor-prone mice—a gain-of-function study as opposed to the many loss-of-function studies done previously.

In most tumor cells, DNA is unusually short on methyl groups. Yet the same cells often contain short sequences replete with methyl groups, hot spots that typically fall on the regulatory regions of genes. After Linhart and Moran methylated these hot spots, the mice developed more macroscopic intestinal tumors than usual.

The pair dug deeper and identified a key growth-control gene affected by the misplaced methyl groups. Their findings, which should be published this spring, provide an interesting twist to the intestinal-tumor tale.

“In these mice, intestinal tumors arise through a complex interplay between genetic events, global hypomethylation and local hypermethylation,” says Linhart, who is still teasing apart the details of this relationship.

The story of the intestinal tumor demonstrates once again that cancer is rarely simple. The term encompasses a multitude of diseases characterized by the abnormal proliferation of cells. Each of these diseases has its own story filled with its own characters, ranging from genes to viruses to methyl groups.

“Epigenetics will not provide a universal cure for cancer because it does not cause every instance of the ‘disease,’” says Linhart. In fact, it might offer more promise as a diagnostic tool. A growing body of evidence suggests that most tumors exhibit epigenetic changes regardless of their origin. So epigenetic patterns could be used to diagnose particular types of cancer, even those caused by genetic mutations. But scientists caution against losing sight of the big picture.

“Although methylation changes can be just as important as mutations in particular cases, epigenetics is just one very narrow part of the broad cancer research field,” Weinberg explains.

Ready for drugs?

Epigenetic marks have attracted attention from pharmaceutical companies hoping to reverse them. In 2004, the Food and Drug Administration approved Vidaza, a DNA-demethylating drug manufactured by Pharmion Corporation, for use in certain blood diseases such as chronic myelomonocytic leukemia. Vidaza is believed to work indirectly by reducing DNA methylation and directly by killing cells.

This approval was the realization of a dream for Peter Jones, director of the University of Southern California/Norris Comprehensive Cancer Center.

Jones was one of the first researchers to observe methylation patterns in cancer cells during the late 1970s and early 1980s. He was also the first to change those patterns by treating cells with chemicals.

One of the chemicals he used was azacytidine—which became Vidaza, 25 years later. Jones believes other success stories will follow.

“I think these drugs will find much more use in the future because they’re very good at resetting the epigenetic program, which has gone awry in a cancer cell,” he says.

But Jaenisch worries that companies will rush to create drugs before fully understanding the consequences of taking them. He cautions scientists to search for side effects before applying epigenetic therapies. This warning comes from experience. When the Jaenisch lab reduced the number of methyl marks on the DNA of tumor-prone mice, the animals developed fewer macroscopic tumors in their intestines. But in another study, the lab found that loss of methyl marks can cause aggressive lymphomas.

“If you want to use methylation changes as a therapeutic tool, you have to know what you’re doing,” says Jaenisch.

Stephen Baylin is more optimistic. He points out that Jaenisch tinkered with methylation patterns in mouse embryos, when enzymes were still busy setting and stripping methyl marks. He wonders if the lymphomas can be blamed on timing, rather than on the treatment itself. Would mice develop these lymphomas if they were exposed to a demethylating drug as adults?

Despite this debate over side effects, Jaenisch agrees that epigenetic therapies will eventually become a reality. “These therapies should materialize after we develop a robust understanding of the mechanisms involved,” he says. “The good news is epigenetic marks are reversible, which gives us hope to treat thousands of cancer patients someday.”

Cancer stem cells and epigenetics

A growing number of scientists accept that not all cells in a tumor are created equal. They believe that a small population of “stem cells” gives rise to the slightly differentiated cells that form the bulk of a tumor. The cancer stem cells divide less frequently than their specialized daughter cells, but live forever.

“It’s still a matter of faith that the stem cell model applies to all cases of cancer, though the evidence is compelling for a small number of solid tumors,” says Whitehead Member Robert Weinberg.

Three recent studies in Nature Genetics, including one by Stephen Baylin, professor of oncology at the Johns Hopkins University, link methylation patterns in cancer cells to patterns of DNA-packaging proteins in embryonic stem cells. The DNA-packaging proteins could leave particular genes, those involved in keeping a cell specialized rather than immature, vulnerable to methylation in adult cells.

“It’s certainly possible that these patterns are fundamentally linked to formation of cancer stem cells, but this needs to be proven,” says Baylin.

Epigenetics and the environment

“The epigenome allows the genome to talk to the environment,” says Whitehead Member Rudolf Jaenisch. In fact, the epigenome might explain the link between particular diets and increased or reduced risk of cancer.

The long-term Harvard Nurses’ Health Study, for example, showed that women who take a multivitamin pill containing folate (a form of vitamin B9) lower their risk of colon cancer by 20 percent. Folate happens to be a methyl group donor, so perhaps it protects the women by acting on the epigenome.

Or perhaps not. “We don’t know if folate modifies the transcriptional state of certain genes, but I do suspect that people have underestimated the plasticity of epigenomes,” says Emma Whitelaw, who studies epigenetics in her lab at the Queensland Institute of Medical Research. She points out that some epigenetic marks in plants fluctuate throughout the day as light levels change. It’s not unreasonable to hypothesize that epigenetic marks fluctuate in humans over a period of days or years in response to diet.

“I think we’re going to discover a lot of layers of epigenetic modification (beyond methylation), and some will be more stable than others,” she says.