Cancer drug developers may be missing their molecular targets—and never knowing it. Many recent drugs take aim at specific cell proteins that drive the growth of tumors. The strategy has had marked successes, such as the leukemia drug Gleevec. But a study now finds that numerous candidate anticancer drugs still kill tumor cells after the genome editor CRISPR was used to eliminate their presumed targets. That suggests the drugs thwart cancer by interacting with different molecules than intended.
The study, published this week in Science Translational Medicine, points to problems with an older lab tool for silencing genes that has been used to identify leads for such drugs. The results also hint that the drugs in question, most of which are in clinical trials, and perhaps others could be optimized to work even better by pinning down their true mechanism.
“The work is very well done and it’s a great public service. I hope people talk about it. I don’t find any of it surprising, unfortunately,” says William Kaelin of the Dana-Farber Cancer Institute in Boston, who has written about why promising preclinical findings are often not reproducible, or fail to lead to drugs.
Leads for many recent targeted drugs emerged from experiments in which cancer cells were dosed with RNA strands that disrupt the natural RNAs that convey a gene’s protein-building instructions. After using this RNA interference (RNAi) method to zero in on genes essential to the growth of cancer cells, researchers screened libraries of molecules to find compounds that block the genes’ proteins.
A few years ago, cancer biologist Jason Sheltzer of Cold Spring Harbor Laboratory in New York and colleagues used CRISPR’s gene-disabling skills, instead of RNAi, to prevent the manufacture of a well-established growth protein, called MELK, in cancer cells. Several companies at the time were developing MELK inhibitors as anticancer agents. But to the group’s surprise, the MELK-deficient cells kept growing. Yet a drug thought to be aimed at MELK still stopped growth of the cells, suggesting its true target was not that protein.
That work spurred Sheltzer’s lab to collect examples of other drugs that target proteins found largely with RNAi. His group ultimately homed in on 10 drugs aimed at six proteins whose roles range from driving cell proliferation to controlling cancer gene activity. When the scientists used CRISPR to knock out the genes for those proteins in various cancer cell lines, the cells kept growing, suggesting the originally RNAi assay was misleading. Yet, when the team gave the relevant drug to cancer cells now missing the target protein, they still died—apparently through some other mechanism. “Many of the previous results were replicable, but the interpretation was wrong,” Sheltzer says.
The researchers found a clue to the real mechanism for a drug, now in preclinical testing, that supposedly blocks a protein called PBK, which aids cell division. By identifying cells that developed resistance to the drug, known as OTS964, and sequencing them for mutations that confer that trait, the lab showed the drug instead blocks the protein CDK11, which plays a different role in cell proliferation. Sheltzer calls this result “exciting” because inhibitors of other CDKs work well against breast cancer, and targeting this one could be a new option. (Science could not reach the drug’s manufacturer, OncoTherapy Science, for comment.)
The developer of a drug on Sheltzer’s list that reportedly activates a protein called caspase-3 that commands cells to self-destruct questioned the study. Chemist Paul Hergenrother of the University of Illinois in Urbana, notes that the drug also activates a related protein with a similar function. So, Sheltzer’s group would have had to knock out the genes for both proteins to eliminate the drug’s effects on cancer cells, Hergenrother says. Another caveat that Sheltzer’s group acknowledges is that some of the drugs’ reported protein targets could influence cancer growth indirectly within the body, for example by spurring other cells to secrete molecules that nurture a growing tumor.
But the new study has made other cancer researchers generally skeptical of RNAi screening to identify potential cancer drugs. That work, it now seems, was “riddled with falsehoods,” says Traver Hart of MD Anderson Cancer Center in Houston, Texas, who now screens for new drug leads with CRISPR. Those results should be checked with multiple techniques because CRISPR can have off-target effects too, Kaelin says. “You have to assume the downstream effects you’re measuring are off-target until you prove otherwise.”
Sheltzer doesn’t think his group’s results cast doubt on the targeted cancer drugs already on the market, as most have other compelling evidence they’re hitting the right protein. But for the 10 candidate drugs studied by his lab, as well as others in development, it’s important to find out how they work so physicians can match patients to the best drug and fulfill the promise of precision medicine, Sheltzer says. Paul Workman of the Institute of Cancer Research in London agrees: “It clearly helps enormously if the true target is now found.”