Silencing cancer

Say “benign” skin cancer, and you get an entirely wrong impression. Familial cylindromatosis is a rare disease that doesn’t kill patients, but does subject them to horrible and painful tumors on the head that must be removed regularly by surgery.

Curiously enough, though, there is hope that this condition can be treated quite successfully with aspirin. There’s a striking little detective story behind this discovery, based on a new and powerful biological tool that Whitehead Fellow Thijn Brummelkamp helped to improve.

The tool is RNA interference (RNAi), and it’s a quick and highly flexible method for letting scientists quickly silence a single gene at a time. It also is highly scalable, which is paying off big time in Brummelkamp’s chosen field of cancer research.

Until recently, our advances in understanding cancer have come with agonizing slowness, with scientists painstakingly isolating genes one at a time and then poking around to find how the genes work together and create a cancer-causing network.

In contrast, Brummelkamp can rapidly set up experiments that screen human cancer cells for thousands of suspect genes. “That way we can quickly find new genes that we suspect may play a role in cancer,” he says. Testing for those genes in samples from cancer patients then offers “a very direct approach to finding cancer treatments.”

Sticking in hairpins

RNAi was first demonstrated in mammals around the year 2000 and is now widely used in labs around the world. The process selectively disables gene expression by attacking messenger RNA, the molecule responsible for delivering the gene’s protein recipe to the cell’s machinery.

Originally, RNA interference was created with “short interfering RNAs,” snippets of RNA about 21 base pairs long. Synthesized chemically, these are expensive and shortlived. In 2002, Brummelkamp and colleagues at the Netherlands Cancer Institute demonstrated a powerful alternative.

The Dutch scientists had noted that the biology of siRNAs was very similar to the biology of microRNAs, the naturally occurring short RNA molecules that also affect gene expression by disabling messenger RNAs. (See “When RNA rules"). Their concept was to get DNA to make hairpin-shaped RNAs quite similar to microRNAs, using common genetic manipulation techniques to do so.

“Other scientists also read the journals, and about 10 groups started to generate a system like this,” Brummelkamp says. “We were the first; that was luck and a bit of hard work.”

They created the hairpin RNAs by using retrovirus vectors (specially created viruses that splice in a segment of DNA). These vectors are easily and cheaply made, replicated and studied with familiar DNA tools. Variations of this technique shot across the biomedical research world; Brummelkamp’s paper is now cited by more than 1,100 other papers.

Pathways to progress

The research that led to the potential cure for familial cylindromatosis began with ubiquitin, a small protein that aids in demolishing proteins whose time is up.

Brummelkamp and his then-co-workers at the Netherlands Cancer Institute knew that enzymes that helped to add ubiquitin to a protein are important in cancer. They speculated that enzymes that aid in removal of ubiquitin also play a major role.

The scientists took 50 human enzymes involved in removing ubiquitin, and used their RNAi toolkit to measure what happened when the gene for each enzyme was silenced on several signaling pathways implicated in cancers. Activity on the NfκB pathway (which stands for nuclear factor kappa B) shot up dramatically when a certain enzyme was silenced.

This enzyme was known to be mutated in familial cylindromatosis—although its function was unknown. The researchers turned to the scientific literature and found that simple compounds such as aspirin inhibit NfκB.

Decoding drugs

After joining Whitehead in late 2004, Brummelkamp is now studying how cancer therapeutics work. That’s not at all as well understood as we might hope. “We may know one target or a few targets, but we don’t typically know the whole biological cascade that a drug uses,” he says.

Postdoctoral researchers Alessio Nencioni and Helen Pickersgill are tackling this problem in tests that silence 7,914 human genes. Studying genes on this scale, “you quickly can see that there are, for example, 20 genes involved in the response of a cell to a particular drug,” says Brummelkamp.

Lab work starts by assembling 23,742 hairpin-RNA retrovirus vectors (three for each of those target genes). Each vector’s gene-specific sequence will later act as a molecular “barcode” for DNA microarray analysis. The vectors are introduced into two dishes filled with human cancer cells, where each hairpin knocks down expression of its target gene. One dish is treated with a drug, the other dish is not.

After leaving the cells in culture for a suitable length of time, the researchers harvest surviving cells from each dish, select and amplify their DNA, label the DNA fragments according to their dish of origin, and plop them on a standard DNA microarray. The relative abundance of each “barcode” quickly indicates genes that make cells more or less sensitive to the drug.

The Brummelkamp lab is now employing this strikingly efficient approach to targeting defective mechanisms in cancer cells. “Cancer cells have found ways of turning off the cellular brakes or jamming on the accelerator to grow uncontrollably,” says Brummelkamp. “But their strengths are also hiding specific weaknesses. We can use RNA interference to find those weaknesses and run them off the road.”

A first such experiment involves drugs that inhibit mdm2, a protein that adds ubiquitin to the p53 protein and thus helps to degrade it. P53 is the king of tumor suppression genes, which is found mutated in about half of human cancers. In the other half of cancers, p53 is not mutated, but the pathway is not functioning well.

Researchers have speculated that in these cancers, inhibiting mdm2 would activate p53, and it would go ahead with its designated role of suppressing the errant cell. If so, that might lead to powerful therapeutics for these diseases. In a paper appearing in April in Nature Chemical Biology, Brummelkamp and several Dutch colleagues identified a gene that helps to explain why Nutlin-3, a small-molecule drug that inhibits mdm2, is surprisingly non-toxic to normal cells.

Moving toward medicine

“If you want to make a conventional drug that inactivates a gene, you can only work with druggable genes,” Brummelkamp notes. “But with RNAi, you can basically inhibit every gene you want.”

Jan Carette, another postdoc in the Brummelkamp lab, is studying ways to inhibit genes that would correct the defect that causes most cases of cystic fibrosis. The disease might be addressed with an RNAi-based inhaler, the researchers speculate.

And labs around the world are delving into other potential RNAi medical applications, studying everything from HIV to, well, hypoallergenic cats. “That’s a nice application of the technology,” Brummelkamp says, smiling. “I’m allergic to cats.”

This article first appeared in the Spring 2006 issue of Paradigm magazine.