Most of the miRNAs they surveyed had lower expression levels in tumors than in normal tissues, irrespective of cell type. But the researchers wondered whether those differences might be due to some other factor, perhaps in the way samples had been collected. To test that, they turned to Tyler Jacks at MIT, whose lab had developed a model of lung cancer in mice. Using precise controls for collecting tissue, they repeated their experiment with the colored bead technique. The mouse studies confirmed that miRNA profiling could distinguish normal cells from cancer cells.

In the second study, Gregory J. Hannon at Cold Spring Harbor Laboratory (CSHL) was naturally drawn to miRNAs through his pioneering work with RNA interference (RNAi)—the pathway that miRNAs use to regulate gene activity—and his longtime interest in cancer biology.

His group built an expression library of all known miRNAs and began their pursuit. “We started screening through miRNAs one at a time to see if any scored as an oncogene in the lymphoma model,” says Hannon. After focusing on a specific cluster of miRNAs, they saw a published paper that connected B cell lymphoma with the chromosomal region that carries the cluster. Meanwhile, Scott M. Hammond, a former postdoc of Hannon's now at University of North Carolina, Chapel Hill, independently found a series of miRNAs consistently overexpressed in a series of human tumor-cell lines. They were part of that same cluster.

This result prompted Hannon and Hammond's teams to collaborate to look at the miRNA cluster in human tumor material across a large set of cancers—B-cell lymphomas and epithelial cancers. They found that the cluster of seven miRNAs, which are produced by a single transcript, is overexpressed most consistently in B cell lymphomas.

To show a direct link between the seven miRNAs and these specific cancers, Hannon and Hammond turned to Scott W. Lowe, also at CSHL, who had been interested in apoptosis control and tumor development since doing graduate work with Tyler Jacks. More recently, Lowe's lab had developed a B cell lymphoma mouse model that they used to characterize several protein-coding genes, including myc, that drive cancer development. Lowe and Hannon together had already tested the model with short hairpin RNAi (essentially synthetic miRNA precursors), showing that some promote tumor development. For the current experiment, the scientists engineered the blood cells of these lymphoma-prone mice to also express the cluster of miRNAs. The mice developed cancer much faster and more frequently.

Hannon sees miRNAs as regulatory molecules, similar to transcription factors. It stands to reason, he says, that certain miRNAs will be relevant to tumor growth. “I would imagine that just like some percentage of [transcription factors], some subset of this class of miRNAs might contribute to cancer as well as other diseases,” he says.

These studies add a new layer of complexity to understanding the origins of cancer, says Jacks, and the labs mentioned above are planning a slew of additional studies. But for now, Jacks says, “We're just now beginning to realize the central role these miRNAs play.”

	PAGE 3 OF 3	Back