Designing drugs that target ‘twin’ cancer proteins

Some proteins in the human body are easy to block with a drug; they have an obvious spot in their structure where a drug can fit, like a key in a lock. But other proteins are more difficult to target, with no clear drug-binding sites.

To design a drug that blocks a cancer-related protein, Scripps Research scientists have taken a hint from the protein’s paralog, or “twin.” Using innovative chemical biology methods, the scientists pinpointed a druggable site on the paralog, and then used that knowledge to characterize drugs that bound to a similar—but more difficult to detect—spot on its twin. Ultimately, they found drugs that only bound to the protein of interest and not its highly similar sibling.

Their approach, described in Nature Chemical Biology on September 18, 2024, and dubbed “paralog hopping,” could uncover new binding sites for drugs and inform drug development more broadly, since nearly half of the proteins in human cells—including many involved in cancer and autoimmune diseases—have such paralogs.

“This method may be generally useful in cases where you have paralogs, and you are trying to find a new drug for one of them,” says senior author Benjamin Cravatt, Ph.D., the Norton B. Gilula Chair in Biology and Chemistry at Scripps Research. “Being able to target one paralog over another is an important goal in drug development, as two paralogs often have different functions.”

Many genes have duplicated throughout evolution, resulting in multiple copies in the human genome. In some cases, copies have evolved slightly different sequences from each other, making their corresponding proteins into paralogs. These protein paralogs remain highly similar in structure and often have redundant or overlapping functions within cells.

In recent years, Cravatt’s research team formulated an approach to develop drugs that bind to the amino acid cysteine—a protein building block with unique, highly reactive chemical properties. The scientists’ method takes advantage of cysteines as an optimal site for drugs to attach to a protein permanently, often inactivating it. However, not all proteins have accessible cysteines. In the case of paralog pairs, one protein may have a druggable cysteine that the other does not.

“We started with this idea that if you know how to drug one protein, you can figure out how to drug its paralog in a similar way,” says Yuanjin Zhang, a graduate student at Scripps Research and first author of the new paper.

As a test case, the team tackled the paralog pair known as CCNE1 and CCNE2. Both proteins have been found to be overactive in breast, ovarian and lung cancer. However, scientists suspected that the two proteins play slightly different roles. The team posited that turning off just one protein could make treating some cancers more effective.

It has been difficult, however, to design drugs that target the CCNE1 and CCNE2 proteins to test this hypothesis. Cravatt, Zhang and their colleagues knew that CCNE2 had a druggable cysteine, while CCNE1 did not. If they could identify drugs that bound to the same spot on CCNE1, even in the absence of a cysteine, they suspected the protein would shut off.

The scientists first engineered a cysteine into CCNE1, mimicking the drug-binding spot they had pinpointed in CCNE2. They then leveraged this neo-cysteine to identify drugs that bind to CCNE1. Next, they screened a library of other chemical compounds for the ability to compete with that drug in binding to CCNE1. The team reasoned that some of the compounds that competed for the same spot would bind in ways that did not rely on the cysteine.

Indeed, Cravatt, Zhang and their colleagues discovered multiple compounds that could bind to the same site on CCNE1 even when the cysteine was removed again. Some compounds did not bind to CCNE2. Some also had opposite functions, stabilizing the molecule so that it might be more active than usual, rather than inactivating it. Structural studies revealed that the CCNE1 compounds bind to a cryptic pocket that was not previously known to be druggable.

The team says the approach highlights the importance of screening for drugs in diverse, creative ways.

“If we had just screened looking for compounds with a particular function, we would not have identified all of these various functional molecules, and if we had just looked at the structure of CCNE1, we would not have found this binding pocket at all,” says Zhang.

More research is needed to discover whether the new compounds have potential utility in treating cancer or other diseases in which CCNE1 plays a role. Next, the scientists plan to apply their paralog-hopping method to other pairs of proteins important for tumorigenesis.

In addition to Cravatt and Zhang, authors of the study include Zhonglin Liu, Sang Joon Won, Divya Bezwada and Bruno Melillo of Scripps; and Marsha Hirschi, Oleg Brodsky, Eric Johnson, Asako Nagata, Matthew D. Petroski, Jaimeen D. Majmudar, Sherry Niessen, Todd VanArsdale, Adam M. Gilbert, Matthew M. Hayward, Al E. Stewart and Andrew R. Nager of Pfizer, Inc.