Mapping proteins to reveal secret drug binding pockets

The human body has over 100,000 different proteins, each made up of a unique sequence of amino acids folded into a complex three-dimensional structure. Figuring out the exact conformation of a protein has long posed a conundrum for scientists, especially when developing small molecule drugs that can bind to proteins and change their activity.

But structures alone only tell part of the story. While a protein’s shape may reveal whether it’s theoretically possible for a small molecule to bind, it doesn’t indicate whether the molecule actually binds in key cellular contexts.

Benjamin Cravatt wears a navy blue shirt against a white background. — For Benjamin Cravatt, chemoproteomics provides a way to do systematic screens of thousands of potential drug-binding pockets.
Credit: Cravatt lab

“Proteins can be modulated in ways that open or close small molecule binding pockets in the cell,” said Benjamin Cravatt, a chemist at the Scripps Research Institute. “We need to identify how small molecules perturb proteins directly in disease-relevant settings.”

So, Cravatt and his colleagues pioneered the field of chemoproteomics 15 years ago to investigate how small molecules interact with proteins in native settings. Now, some of the first candidates of large-scale chemoproteomics screens are entering the clinical pipeline with ambitious targets that pharmaceutical companies have not historically pursued. Companies such as Vividion Therapeutics, co-founded by Cravatt, hope that their unconventional chemoproteomics approach can open new avenues for drug discovery.

Why undruggable proteins remain a challenge

For scientists, “certain categories of proteins have been historically challenging,” Cravatt said. These so-called undruggable proteins, which make up as much as 80 percent of the human proteome, have complex structures that lack easily accessible binding pockets for small molecules. They include many important cellular regulators such as scaffold proteins, transcription factors, and RNA-binding proteins. Without being able to target these key players — many of which are drivers of different cancers, immune conditions, and other diseases — patients can go untreated.

These proteins may not have conventional binding sites, but sometimes they have cryptic binding sites: pockets that can only be detected once a small molecule actually binds. “A lot of secret pockets exist on human proteins,” said Qing Yu, a biochemist at the University of Massachusetts Chan Medical School. “Identifying this pocket allows us to rationally design drugs to inhibit the activity of the protein.”

To find these pockets, chemists could try to define the crystal structure of the protein in complex with the small molecule, but this is a slow, tedious process that is not tractable for many proteins. Instead, about a decade ago, Keriann Backus , then a postdoctoral fellow in Cravatt’s lab, began thinking about more systematic approaches. At the time, some of the most intriguing drugs were covalent kinase inhibitors for cancer such as ibrutinib and afatinib. These molecules made strong, lasting bonds with cysteine amino acids in their target protein to inactivate key phosphorylating domains, and researchers had shown these drugs to be safe and efficacious in the clinic. “Covalency was a modality that was potentially very impactful,” Backus said.

Yuanjin Zhang and Rachel Hayward wear white lab coats while working in a laboratory. — Yuanjin Zhang and Rachel Hayward worked in the Cravatt lab to map covalent binding sites on proteins by using chemical probes.
Credit: Cravatt lab

Moreover, the researchers realized that covalent drugs could be especially useful for finding cryptic binding sites. “If there's a dynamic movement in a protein to create a pocket that's transiently open, covalent chemistry can trap proteins in their open, druggable state to have a much more tight binding,” Cravatt said.

Backus set out to find a way to map all the cysteines in a protein where a small molecule ligand could potentially bind. To do this, she created a library of small molecule fragments attached to charged chemical tags and treated cell lysates with individual fragments from the library. Proteins in these cell lysates were in the natural state that they would be found in in a cell — complete with other binding partners, modifications, and state-dependent configurations — so the fragment could bind in whatever binding pockets it could find on these proteins.

Then, Backus coated the proteins with a probe designed to bind to cysteines — but only able to bind accessible cysteines that were not already bound by a small molecule. By using mass spectrometry to compare proteins with and without the small molecule, she identified which cysteines in the protein were bound by small molecules to identify the pocket where the small molecule could bind. She published this method in Nature in 2016 (1).

This “pioneering work,” according to Yu, who was not involved in the study, opened the door to comprehensively profiling reactive cysteines in proteins in different cellular states or perturbations. At first, Cravatt and Backus weren’t sure how valuable that would be. “However, so many of the fragment ligandability sites that we found occurred on cysteines that had not been known previously to be targeted by small molecules,” Cravatt said. “We were opening up potentially new druggable space in the proteome.”

Scaling chemoproteomics from the lab to biotech startups

To Cravatt, one strength of the approach was its ability to scan thousands of sites using a targeted library of small molecule fragments. But he knew there was room to scale it up further — with some help.

In 2017, Cravatt co-founded the biotech startup Vividion Therapeutics alongside two fellow chemists from Scripps Research Institute, Phil Baran and Jin-Quan Yu. The company’s goal was to use the new system developed by Backus to do large-scale screens and identify potential small molecules to develop into drug candidates.

We were opening up potentially new druggable space in the proteome.
- Benjamin Cravatt, Scripps Research Institute

“Vividion did an exceptional job of scaling up the platform so that they could look at many more compounds against sites of interest in a suitable time frame,” Cravatt said.

Scaling the platform has paid off. The company is now pursuing targets for a variety of different diseases, including four disclosed targets for cancer. Some of these molecules target popular cancer pathways. For example, one Vividion Therapeutics candidate, VVD-130850, is a small molecule that targets the STAT3 transcription factor in solid and blood cancers and keeps it from binding DNA to regulate the expression of certain genes. The oral drug entered a Phase 1 clinical trial last year. Another candidate, VVD-159642, targets the RAS-PIK3CA pathway, a cell proliferation cascade known to drive many solid tumors. The team dosed their first patient in a Phase 1 clinical trial testing this drug in April.

Vividion Therapeutics is also pursuing candidates with unique mechanisms of action. In 2024, they published a study in Nature describing a new molecule to target the Werner syndrome (WRN) helicase, a DNA repair protein that can cause cancer when defective (2). Malfunctioning WRN helicase creates microsatellite instability, short sequences of repeating DNA that become misaligned during DNA replication and can cause cancer.

“This molecule is hitting a major target that people have been chasing for a long time,” said Backus, now a chemist at the University of California, Los Angeles who is not associated with Vividion Therapeutics.

As a postdoctoral fellow, Keriann Backus developed the technology now being used by Vividion to identify drug candidates. Now, she hopes her lab’s work can make these screens more scalable and effective.

Credit: Penny Jenning

The molecule that the Vividion Therapeutics team discovered through their screen binds covalently to the WRN helicase away from its active site and causes a structural change to the helicase that prevents it from binding to DNA, known as allosteric inhibition. In their study, the team showed that the drug could stop cell division and kill cell lines that had microsatellite instability, while sparing healthy cell lines. With promising findings that the drug caused tumor regression in patient-derived xenografts, the company is partnering with Roche to test it in a Phase 1 clinical trial.

“They are really showing the diversity of how small molecules can regulate protein function through allostery, which is super neat,” Cravatt said.

Yu was particularly impressed by how this approach could identify ways to inhibit specific members of a protein family that all have similar structures. For example, in 2022, Vividion Therapeutics researchers published a study describing a molecule that can target JAK1, a key protein in an immune signaling pathway (3). JAK1 is part of a family of proteins that all have very similar structures. Conventional drug discovery to target JAKs has typically focused on the active site of the protein. However, that is the portion of the protein that is especially conserved across members of the JAK family, so most JAK inhibitors developed so far have targeted all members of the family. This results in broad inhibition of the JAK protein family, which can cause side effects.

This molecule is hitting a major target that people have been chasing for a long time.
- Keriann Backus, University of California, Los Angeles

By instead doing a protein-wide search for reactive cysteines, the Vividion Therapeutics team found a cysteine outside of the active site that is specific to JAK1 but could still influence the protein’s activity.

To Yu, this is the unique pathway to drug discovery that chemoproteomics enables. “Imagine you can use this approach generally on all the kinases, not focusing on the kinase domain, but on the allosteric sites,” he said. “You can have selective kinase inhibitors, which has been a long-time goal for many scientists.”

Expanding beyond cysteines to map new drug targets

Improvements in mass spectrometry over the last decade have made it possible to measure much rarer and less abundant proteins and to increase the amount of data collected from a sample. But there’s still room for improvement. “We've only been able to access maybe 20 to 30 percent of the reactive cysteine in the human proteome,” Yu said. “There's a huge space that we haven't explored yet.” That’s not even accounting for the other 19 amino acids in the proteome. Researchers so far have focused on cysteines because of their potential for covalent bonding, but Yu noted that new chemistry that allows screening other amino acids could unlock additional binding pockets.

In Backus’ lab, she hopes to further improve the technology by developing better reagents for chemoproteomics experiments. By improving sample cleaning and processing, she hopes that the screens will work better and enable deeper insights into the proteome. Meanwhile, Yu’s group is trying to make chemoproteomic screens more scalable to test thousands of compounds at once. He has developed a method to tag samples with a barcode and combine more than 30 samples together for one analysis.

As the technology improves, companies are starting to invest. Backus said she is aware of at least 10 startups using chemoproteomics for drug discovery. At the same time, scientists are also redefining the kinds of questions they can ask using chemoproteomics. “The field is still figuring out the target space,” Backus said. “It all started with covalent kinase inhibitors and is broadening from there.”

Qing Yu smiles while wearing black-rimmed glasses inside against a white background. — Qing Yu suspects that expanding chemoproteomic screens beyond cysteines could reveal even more binding pockets.
Credit: Qing Yu

For example, chemoproteomics could enable the long-sought-after ability to target cryptic binding sites on transcription factors that appear only when they interact with DNA and other cofactors. In 2024, Cravatt’s group published a study using their chemoproteomic screen to identify covalent modulators of the FOXA1 transcription factor (4). This transcription factor plays an important role in reshaping DNA for gene regulation, and they identified a small molecule that changes how the transcription factor interacts with DNA in cancer cells. “Doing the screen in cells helped us find a kind of state-dependent, DNA-dependent, liganding event,” Cravatt said.

One goal that’s still on the horizon is to incorporate function into chemoproteomic screens. Currently, the screens can identify small molecules that bind to the target protein, but they don’t show whether that binding affects the function of the protein — ascertaining that requires additional experiments. Recent work from the Cravatt lab has shown that CRISPR or protein interaction assays can be integrated into chemoproteomic screens to test for functional consequences alongside binding (5,6).

Ultimately, not every candidate from a chemoproteomic screen might be a viable drug — and to Yu, that isn’t necessarily a problem. “We can use [the molecule] to study the protein’s function, or the drug discovery industry can take the candidate into later stages to make a real drug,” he said. “Either way, it's going to be really valuable to the community.”

Frequently asked questions (FAQ)

What is chemoproteomics in drug discovery?

Chemoproteomics is a technique that uses chemical probes and mass spectrometry to map how small molecules bind to proteins inside cells. This helps researchers identify hidden or cryptic drug binding pockets that can’t be seen through protein structure alone. By revealing these sites, chemoproteomics expands the number of druggable proteins in the human proteome.

Why are some proteins considered “undruggable”?

Proteins are often called “undruggable” when they lack clear binding pockets for small molecule drugs. These include transcription factors, scaffold proteins, and RNA-binding proteins — many of which are central to diseases like cancer and immune disorders. Chemoproteomics is making it possible to target these proteins by finding cryptic sites that only appear in certain cellular states.

How does chemoproteomics find hidden binding pockets?

Chemoproteomic methods use chemical fragments that bind to amino acids such as cysteines in proteins. By tracking where these molecules attach in live or native conditions, researchers can reveal transient binding pockets that traditional protein structure studies might miss. This allows for the discovery of new drug targets once thought inaccessible.

What diseases could benefit from chemoproteomics-based drugs?

Many early drug candidates discovered through chemoproteomics focus on cancer, particularly solid tumors and blood cancers driven by undruggable proteins like STAT3 or RAS-PIK3CA. Researchers are also exploring applications in immune disorders and rare genetic conditions. As the technology matures, chemoproteomics could accelerate drug discovery across a wide range of diseases.

What is the future of chemoproteomics?

Future directions include expanding beyond cysteine reactivity to other amino acids, integrating functional assays like CRISPR, and scaling screens to cover thousands of proteins at once. These advances could open entirely new drug discovery pathways, helping scientists develop precision medicines for currently untreatable diseases.

References