11/4/2005

Science

By Jennifer Couzin

MicroRNAs are being implicated in various human cancers, and scientists are trying to sort out just how culpable they are.

For Frank Slack, the story began when his worms exploded through their vulvas.

It was 1997, and the developmental biologist, now at Yale, had been tinkering with microRNAs (miRNAs), tiny RNA molecules that regulate gene expression. Slack is a worm man, and in his wriggly subjects he had deleted the gene for just one of the 120 known worm miRNAs.

The developing animals' stem cells failed to morph into specialized cells as they normally do and instead kept dividing. "The worms looked extended, weirdly floppy; they kind of looked uncoordinated," he says. The vulvas didn't develop properly and ruptured. A worm skeleton is under hydrostatic pressure, and with the rupture, "the animals burst through," an experience that killed roughly half of them.

When Slack probed the underlying genetics, he uncovered something tantalizing that linked these unfortunate animals to human biology. Deletion of this miRNA, called let-7, prompted overexpression of a gene, Ras, that's strongly associated with many cancers. In other words, when let-7 is expressed normally, it seemed, it blunts Ras. Since Slack's find, the let-7-Ras story has unfolded rapidly, one of a growing bundle of strands tying miRNAs to cancer.

More than a dozen papers have shown that miRNAs are expressed differently in cancerous tissue. Braided together, the latest miRNA discoveries suggest potentially vast roles for the tiny molecules in malignancy; they have also sparked spirited debate over whether miRNAs are driving cancer or are simply a marker of it. Either way, the nascent field could eventually assist doctors in diagnosis, prognosis, and possibly treatment. Last week, a paper in the New England Journal of Medicine (NEJM) reported that 13 miRNAs form a signature associated with prognosis and disease progression in patients with chronic lymphocytic leukemia (CLL), a cancer of blood.

Sorry fate. A worm without a microRNA bursts through its vulva; replacing the microRNA keeps the worm intact.

"There is a whole other world out there, which I don't think we know anything about," says Phillip Sharp of the Massachusetts Institute of Technology (MIT) in Cambridge, who has studied small RNA molecules for years and is examining their influence on tumors.

Cancer Connections

With rare exceptions, it's far from clear which genes the miRNAs are targeting, how many miRNAs are involved in cancer--and how they're involved--and what governs miRNA behavior. Uncertainties aside, however, Sharp and others are not surprised that miRNAs are being implicated. Many of the dozen or so animal miRNAs of known function play a big role in early development. In fruit flies, some miRNAs govern apoptosis, or cell death; in worms, as Slack witnessed to dramatic effect, they control cell differentiation. Both processes, like many others in development, are critical components of tumor formation and spread. "There were these clues," says Joshua Mendell, a geneticist and molecular biologist at Johns Hopkins University in Baltimore, Maryland, who set up his own lab last year and began exploring the miRNA-cancer connection.

Mendell chose to focus on a proto-oncogene called c-Myc; proto-oncogenes (Ras is another) are often highly expressed in cancerous tissue and implicated in initiating malignancy. "Even though Myc has been studied for several decades, [it's] still not fully understood how it causes tumors," says Mendell. Examining a human cell line in which c-Myc expression could be manipulated, Mendell and his colleagues found that when expressed, c-Myc activates a cluster of six miRNAs. More important, another gene that's both a target of c-Myc and drives cell division damps down its expression when two miRNAs in Mendell's cluster are active. That suggested that this miRNA pair could control the balance of cell death and proliferation driven by c-Myc.

While Mendell and his team were sifting through their cell samples, a cell biologist at the University of North Carolina, Chapel Hill, was studying how miRNAs might drive lymphoma. Unaware of Mendell's findings, Scott Hammond hit on seven relevant miRNAs in human cancer cells; the cluster was nearly identical to Mendell's list. "We both kind of came to the same group of miRNAs," says Hammond.

But Hammond recognized a problem. Cancerous cells contain abundant abnormalities, many a result of cancer rather than a cause. Hammond didn't know into which category his miRNAs, which were strikingly overabundant in cancer tissue compared with normal tissue, fell.

Teaming up with Greg Hannon at Cold Spring Harbor Laboratory in New York, the pair and colleagues forced overexpression of six of the miRNAs together in 14 mice predisposed to a form of lymphoma. Cancer accelerated dramatically. After 100 days, all the treated mice had cancer, compared with about a quarter of controls. The work is "precedent setting," says Sharp, one of the first times miRNAs have been shown to spark cancer. If other miRNAs are found to target either proto-oncogenes, which can trigger cancer, or tumor suppressors, which squelch it, that would further incriminate them.

Hammond and Hannon's work appeared in Nature this past June, along with the studies from Mendell's lab and from Todd Golub of Harvard Medical School and the Dana-Farber Cancer Institute in Boston and his colleagues. Golub's research used expression of miRNAs to classify different types of tumors.

But the Hammond-Hannon work remains the exception; nearly all the research implicating miRNAs in cancer does so indirectly. One of the only other studies showing potential causality comes from Carlo Croce of Ohio State University in Columbus, the first cancer geneticist to publish on miRNAs. In September, Croce reported that in patients with CLL, the loss of two miRNAs boosts expression of a gene promoting cell survival. The gene is believed to help drive the leukemia in its earliest stages. Without the miRNAs that mediate it, leukemia can set in.

Elusive Quarry

In retrospect, says Harvard RNA expert Gary Ruvkun, given the broad roles being assigned to miRNAs in cancer, it's amazing that cancer geneticists so thoroughly missed miRNAs. "I just find it hard to believe that the cancer people were that lame," says Ruvkun, who is just now starting to back a miRNA-cancer connection.

"We share a collective guilt as a community," agrees Ren Bernards, a cancer geneticist at the Netherlands Cancer Institute in Amsterdam who is not studying miRNAs. With a laugh, he recalls his graduate school days, when he tossed "anything small, degraded, uninteresting" in the trash. At the time, miRNAs fell squarely in that category. Furthermore, miRNAs are generated by genes that don't produce proteins--long derided as "junk" DNA.

Indeed, Croce, now a consummate miRNA fan, admits being dragged into the field unwittingly. Ten years ago, he grew convinced that a CLL tumor-suppressor gene was nestled in a certain stretch of DNA--but he couldn't spot it. Baffled and stubbornly determined, Croce turned to colleagues in the CLL field, who handed over additional leukemia samples to scour. Only when Croce stopped looking for a coding gene 3 years ago did he settle on the two miRNA genes he's been studying ever since.

With the outlines of a miRNA-cancer connection taking shape, researchers are now beginning to tackle some of the toughest questions. Perhaps the most vexing involves finding miRNA targets. Like other types of small RNA molecules, miRNAs influence genes with a similar sequence--but the match need not be exact, making the targets maddeningly hard to pin down.

Rainbow dysfunction. Profiles of 218 tumor samples from various cancers show miRNA expression as colored "hills."

No experiment "can hand you a target on a silver plate," says Nikolaus Rajewsky, a biologist and mathematician at New York University. These days, says Rajewsky, the best target-finding melds two tactics. The more traditional compares putative miRNA targets in mammals with known targets for the same miRNA in other species. The other calls for over- or underexpressing a miRNA, then running microarray studies to spot affected genes. But "the computational approaches are still evolving; the experimental approaches are labor-intensive," says Victor Ambros, a geneticist at Dartmouth Medical School in Hanover, New Hampshire. "What we're not sure about is how many targets we're missing."

Several labs are conducting massive miRNA knockout studies to delineate the targets and functions of individual miRNAs. At the University of California, San Francisco, RNA biologist Michael McManus is leading a six-person mouse miRNA consortium; it plans to delete each of the 350 known miRNAs in mice, one at a time.

Under the influence. Chronic lymphocytic leukemia cells (above) appear to be driven by miRNAs.

But in cancer especially, biologists warn, painting a comprehensive miRNA picture will likely be exceedingly complex. When miRNAs "get overexpressed or underexpressed or deleted, lots of things can happen," says Tyler Jacks, director of MIT's Center for Cancer Research. "And trying to figure out exactly which of those things is contributing to tumorigenesis or prognosis or what have you" calls for "a lot of detective work."

Nor is it clear what prompts miRNAs to misbehave in the first place. "We'd really like to know," says Slack, who theorizes that mutations in miRNAs could be at fault, as could defects in transcription factors, proteins that control gene expression. Croce has found two leukemia patients born with the miRNA mutations implicated in CLL.

Looking Ahead

Given all the unknowns, miRNAs are a long way from the clinic. But some drug companies are dabbling in them nonetheless. Jan Weiler, a chemist at Novartis in Basel, Switzerland, has been studying the role of miRNAs in disease for 2 years. (In addition to cancer, the molecules are tentatively linked to neurological disorders and diabetes.) "It's a lot of speculation, a lot of hope," says Weiler, who envisions perhaps delivering miRNAs to patients lacking them. "If we don't look at it now, we're probably too late," he says, while acknowledging the risk that "maybe … in 3 years' time, the whole thing is dropped."

If therapeutics remain distant, diagnostics are closer to reality. Croce co-authored last week's NEJM paper that reported on a 13-miRNA signature in CLL. His group also found that among 94 CLL patients, many of those lacking Croce's original two miRNAs have a milder form of CLL, whereas most with the two functioning miRNAs suffer a more aggressive form. "It looks like CLL is not one disease but two," he says, and the distinction could be useful in diagnosing and treating the leukemia.

Other cancers, too, are being eyed as harboring miRNA culprits. One of the very first miRNAs tied to cancer--let-7 with its exploding worms--was last year found to be lacking in lung cancer tissue taken from patients in Japan. Those with the lowest levels fared the worst--suggesting once again that flawed miRNA expression bodes poorly for one's health.

Copyright © 2005 by the American Association for the Advancement of Science