"People have this bias that the majority of gene expression control is at the transcriptional level," Mariano Garcia-Blanco, professor of molecular genetics and microbiology at Duke University Medical Center, told BioWorld Today. "That's absolutely wrong."

Garcia-Blanco studies another way that cells can modify their gene expression: through alternative splicing, the process by which the exons of a gene are stitched together in different ways to increase gene expression variability. And in a paper to be published in the Sept. 19, 2006, issue of the Proceedings of the National Academy of Sciences, now available online, he and his colleagues published a reporter system for observing alternative splicing in vivo that yielded new findings about the nature and significance of cell-type transitions in cancer metastases.

The epithelial-mesenchymal transition, or EMT, is a change in cellular characteristics that is prevalent during development and also thought to be an important step in metastasis. In an epithelial-type state, cells are fairly tightly embedded in an extracellular matrix. When they transition to the mesenchymal state, they become unmoored from their surroundings, which is what allows them to enter the bloodstream and form metastases.

But Garcia-Blanco and his colleagues, whose main goal at the time was to develop a reporter system that would allow them to visualize splicing variants in vivo, found that their system allowed them to see mesenchymal-to-epithelial transitions. While the epithelial-to-mesenchymal transition is a well-known phenomenon in cancer cells, the opposite transition "had been postulated but not observed," Garcia-Blanco said.

The scientists were using their imaging method to study alternative splicing of the fibroblast growth factor receptor 2 in two cell lines derived from a type of rat prostate tumor. Exon IIIc of the receptor is expressed in one type of tumor cells, AT tumors, which consist of mesenchymal-type cells; another type of tumor cells, DT cells, expresses exon IIIb - which, according to Garcia-Blanco, makes a "huge" difference.

"It's the same receptor, but when you alternatively splice it, you completely change the ligand that it binds," he said. DT cells usually have epithelial characteristics.

The researchers constructed a reporter gene that would turn cells fluorescent when exon IIIb was included in cells, but not when alternative splicing included exon IIIc instead. Cell culture experiments confirmed that tumor cells with exon IIIc were not fluorescent, which emboldened Garcia-Blanco and his team to try their method in vivo. And to their surprise, while most cells in the AT tumors expressed mesenchymal-type exon IIIc and so were not fluorescent, the researchers found clusters of fluorescent cells, which must be splicing epithelial-type exon IIIb into the fibroblast growth factor, in the majority of the tumors they looked at.

Staining for another epithelial-cell marker confirmed that those cells were indeed transforming from mesenchymal to epithelial-type cells; such clusters mostly were found near stromal cells, which are tumor-support cells. The scientists found fluorescent cell clusters in lung metastases. One possible explanation is that once they reach a new destination, stroma somehow may induce mesenchymal-type tumor cells to revert to an epithelial type, enabling them to settle down in their new digs.

Garcia-Blanco said that the results have implications for how to conceptualize cancer in general; instead of the ability to attain one specific state, such as a mesenchymal-like state, what may make cancer cells dangerous is the ease with which they switch between different states - or, as he phrased it, "Perhaps the most important thing is, 'How many hats can you wear?'"

On an applied level, Garcia-Blanco also hopes that preventing cancer cells from transitioning might be a way to selectively target cancer cells in the clinic. He cautioned that there are other cellular phenomena that have been observed in cancer cells before they were found in normal cells. But, he said, once development is over, switching back and forth between different states "could be a very specific property of cancer cells, whereas cell growth is definitely not."

"That's our hope," he said. "And you can bet we are working very hard right now to figure out whether that's true."