Molecular Glues: The Next Frontier in Targeted Protein Degradation

Unlocking a new class of medicines

For decades, drug discovery has focused on inhibiting enzymes to treat disease. Yet vast swathes of the proteome remain “undruggable” by traditional approaches. Molecular glues – small molecules that induce two proteins to bind and trigger targeted degradation – offer a way to remove harmful proteins from cells entirely.

Some in the field of drug discovery, like Dr. Ben Cross, believe these molecular glues could transform the whole landscape of disease treatment.

Cross is the chief technology officer at PhoreMost, a pre-clinical drug discovery company based in Cambridge, UK.

“These molecules have a fantastic promise and very, very dynamic potential to target any given disease,” he told the Technology Networks audience during “Innovations in Biopharma 2025”.

“That’s why I very boldly titled my talk ‘Molecular Glues to Drug Everything.’”

The ubiquitin–proteasome system: Nature’s recycling bin

Cells maintain protein homeostasis, or “proteostasis,” through the ubiquitin–proteasome system. Proteins destined for disposal are tagged with ubiquitin and fed into the proteasome – “often called the garbage can of the cell,” Cross explained. While proteasome inhibitors are already used clinically, the system can also be co-opted to eliminate disease-causing proteins.

Targeted protein degradation harnesses this machinery. The first generation of degraders, proteolysis-targeting chimeras (PROTACs), are large, bivalent molecules that bind both a target protein and an E3 ligase. PROTACs have delivered “phenomenal data” in early trials, but their size can cause distribution, toxicity and resistance challenges.

Molecular glues: Small but mighty

Molecular glues achieve the same end – degradation – without the bulk. They are monovalent small molecules that subtly reshape an E3 ligase’s surface to promote binding to a target protein. Many were discovered serendipitously, often during clinical use for unrelated purposes, and later recognized as glues after mechanistic studies.

The canonical example is CC-885, which binds cereblon (CRBN) and induces degradation of GSPT1, an essential protein in cancer biology.

“It’s clear that we need much more diversified approaches to discovering these kinds of drugs,” Cross said during his presentation. “Nobody has come up with a rationalized or definitive, systematic way to discover molecular glues as of today.”

The discovery challenge

Unlike PROTACs, molecular glues cannot be designed by simply linking two known ligands. Predicting the induced protein–protein interactions they require is complex – even with today’s advances in structural prediction. High-throughput brute-force screening is difficult because of the vast number of possible protein pairs and binding modes.

As Cross noted, “We’re still just starting to understand the rules… actually predicting how proteins come together in cells and how they come together in a way which is functionally relevant is not always that trivial.”

PhoreMost’s GLUESEEKER platform

To tackle this, PhoreMost has developed GLUESEEKER, a platform that inverts the problem. Instead of starting with chemistry, it begins by inducing protein–protein interactions directly through engineered mutations, then uses those as blueprints for small-molecule design.

“We are generating vast libraries of edited E3 ligase proteins,” Cross explained. “We ask: can we induce two proteins to come together in a way which is productive for their degradation? And does that interaction have a property we can subsequently drug?”

The team focuses on inserting short loops of amino acids into specific sites on an E3 ligase to create a “neo-surface” for target binding. These engineered ligases are screened in cells for their ability to degrade a protein of interest: for example, GSPT1.

From engineered loops to drug candidates

One key proof-of-concept came when the team inserted just three amino acids into cereblon, enabling it to degrade GSPT1 with high specificity. This effect was confirmed biochemically, with the engineered ligase forming a ternary complex with GSPT1 only in the presence of pomalidomide, an existing cereblon ligand.

Crucially, the loop insertion’s structure provided a pharmacophore – a 3D arrangement of chemical features – that could be mimicked in silico. Computational screening of millions of small molecules identified candidates predicted to replicate the loop’s binding effect on wild-type cereblon. When tested, some of these compounds directly induced degradation of GSPT1 in cells, matching the potency of CC-885.

“These came out of a computer, into a test tube, and then into some cells, and straight away functioned as molecular glue degraders,” said Cross. “That, to us, is super exciting […] it’s incredibly rapid.”

Expanding the toolkit

PhoreMost has since broadened GLUESEEKER to other E3 ligases, with early work targeting undruggable oncogenes such as STAT3 and β-catenin, as well as neurodegeneration-linked α-synuclein. Because the platform identifies functional protein–protein interactions in living cells, it is well-suited to challenging targets like intrinsically disordered proteins, which are hard to study in vitro.

Changing the Landscape of Screening: AI-Powered ASMS Ligand Identification

This webinar explores how advanced screening technologies strengthen affinity selection mass spectrometry (ASMS) workflows to speed up timelines, improve precision and uncover novel binders.
View Webinar

Advertisement

The approach is also portable: “The degree to which this technology is portable to any ligase or any product of interest is really readily apparent to us,” Cross told the audience.

Overcoming resistance and diversifying ligase use

One of the drivers for finding new ligases is resistance. Cancer cells treated with cereblon-based degraders can simply downregulate cereblon, rendering the drug ineffective. Diversifying the E3 ligases used – and the ways they can be modulated – is therefore key to durability.

Cross is optimistic: “There are probably going to be enzymes which are highly conducive or permissive for these kinds of protein–protein interactions […] we just aren’t advanced enough in the field yet to be able to predict that without doing these kinds of screens.”

A step toward systematic glue discovery

By turning induced proximity into a measurable, engineerable property, GLUESEEKER offers a systematic route to molecular glue discovery – potentially removing much of the serendipity that has defined the field to date.

In Cross’s view, the goal is ambitious but attainable: “We want to do things with molecular glues that have never been done before […] to generate new medicines for diseases which have been historically intractable.”

This content includes text that has been generated with the assistance of AI. Technology Networks' AI policy can be found here.