By David N. Leff

In northern Alaska, southern Greenland, Siberia and Iceland, people frequently witness the multicolor fireworks extravaganzas known as Northern Lights, or the aurora borealis.

Fruit-fly geneticist David Glover observed the light show through a microscope. What he saw, actually, was a radial display in a mutated gene product that had gummed up its cell-division works. Specifically, the mutant protein prevented separation of its cell's bundle of chromosomes. This led to the formation of one mitotic spindle instead of two.

That truncated array of unipolar microtubules, when stained, reminded Glover of a Northern Lights display, so he named the gene aurora. That was in 1995.

This year, when molecular biologist Gregory Plowman and his collaborators discovered two human genes that were genomic look-alikes for Glover's Drosophila aurora, he named them aurora-1 and aurora-2. The latter sequence in particular showed salient earmarks of a new and potent oncogene, as reported in the European Molecular Biology Organization (EMBO) Journal, dated June 1, 1998.

Plowman, vice president of molecular biology at Sugen Inc., in Redwood City, Calif., is that EMBO paper's senior author.

So far, he and his co-authors have located both auroras to the cell's mitotic apparatus. (Mitosis is the genetic process that orchestrates the machinery by which a parent cell divvies up its DNA legacy between its two daughter cells.)

"Aurora-2," Plowman told BioWorld Today, "is located at the spindle poles that are pulling the chromosomes apart. The other protein, Aurora-1, is located at the equatorial plate, where the two daughter cells are budding apart.

"That's at the cell-division stage of cytokinesis," Plowman explained. "That's the final step of mitosis, when one cell becomes two. It's at that final junction that aurora-1 is expressed."

In normal cells, aurora-2 seems to be responsible for a kind of nanny job, making sure that each newly divided daughter cell is equipped with the correct number of chromosomes. "This chromosome segregation is often upset in cancer cells," Plowman pointed out, "with possibly aneuploidy chaos — a random shuffling of chromosomes."

The human aurora-2 gene maps to the long arm of chrosomsome 20. This is a hot-spot region, amplified in a variety of cancers, notably colorectal.

The Sugen scientists found that in 24 out of 25 tumor types (96 percent) they surveyed, the aurora-2 gene was overexpressed. "So then we did the classic studies," Plowman recounted. "We asked: What are the effects of overexpressing the gene in some cultured cell lines? In vitro and in mouse tumors, we found that it was transforming the cells — acting like an oncogene.

Overexpression + Amplification = Cell Transformation

"The bottom line," he added, "is that an overexpressed gene leads to a cascade of more RNA, more protein, more activity. And more activity results in malignant transformation."

What's more, the team found that colorectal cancer cells contained enormously amplified aurora-2 copy numbers. They discovered this gene amplification in 41 out of 79 (52 percent) fresh colon carcinoma biopsies.

"Amplified genes are just not very common," Plowman pointed out. "Especially not genes that are as prevalent as aurora-2. Here we're talking about a gene that's multiplying in up to 52 percent of colon cancers."

He also emphasized that the homology of aurora-2 to the fruit fly version "suggested that it is involved in the process of segregating chromosomes. Not only that, but it was at a place apparently involved in regulating the microtubule architecture of the cell itself. That's a target often used for cancer chemotherapy, specifically paclitaxel. So we wanted to know: Is this a possible tumor target?"

The co-authors' finding that indeed it is "means a couple of things to Sugen," Plowman observed. "One is that we finally have a target in colon cancer that's a positive regulator of tumor growth. Instead of a bunch of excellent and informative tumor suppressor genes that are diagnostic reagents for studying familial cancer syndromes, this is one that appears to be implicated in possibly half the cases of colon carcinoma.

"That's what drug companies look for," he continued. "And we like aurora-2 because it shows a selectivity by being amplified in the tumors."

Tumor Cells Gatecrash Cell-Cycle Checkpoint

The aurora-2 gene, the co-authors report, encodes a serine-threonine kinase enzyme. Aurora-2 gene amplification increases its activity in cell division. "That intracellular kinase signal is still being defined," Plowman said. "We believe the message is one that tells the dividing cell: 'Everything is O.K. Go ahead with pulling the chromosomes apart.' If this signal isn't met during mitosis, everything would stop, at a checkpoint in the cell cycle progression. In a tumor cell, that disruption would cause it to push through that checkpoint gate, when the cells aren't ready."

Plowman explained, "What we do in our whole drug-discovery platform at Sugen is focus on the enzymes — mainly the kinases. Then we make a high-throughput screen that can handle thousands of compounds a day, and apply small-molecule libraries to them. Ultimately, chemistry will be done around those small molecules, and at that stage they go into animal trials.

"Most of our in vivo work," he went on, "will be done with small-molecule aurora-2 inhibitors, once we have selected those. We've already done some stuff on them, but not a lot."

Plowman pictures this eventual candidate anti-colon-cancer compound as an anti-mitotic drug. "I think its application," he said, "would be similar to paclitaxel, but possibly without as many toxic side effects. Paclitaxel," he pointed out, "is also anti-mitotic, because it disrupts the dividing cell's microtubules."

He continued: "Our whole hope is that there will be enough of a therapeutic window to affect rapidly proliferating cells to some degree. But I think that because of its amplification, it's more likely to hit the tumor cells more selectively."

He supposes that "we can get an inhibitor molecule identified, do the chemistry, and turn that lead into an actual drug within the next three to four years. Then clinical trials will take anywhere from two to four years."

Plowman concluded: "I think this is a good tumor target, an incredibly good screen. It will depend a little bit on luck, and a little on innovation, I guess." *