Targeted protein degradation (TPD) is a strategy used to selectively remove proteins from cells, rather than simply blocking their activity. Traditional small-molecule drugs work by binding to a protein and inhibiting its function, leaving the protein intact. In contrast, TPD harnesses the cell waste-disposal system—in particular, the ubiquitin-proteasome pathway—to tag the target protein for destruction. Once tagged, the protein is chopped up and recycled by the proteasome, eliminating it from the cell.
Perhaps the best known TPD approach uses PROTACs (proteolysis-targeting chimeras), which are bifunctional molecules: one end binds the protein of interest, and the other recruits an E3 ubiquitin ligase. By bringing the protein and ligase together, the PROTAC triggers ubiquitin tagging and subsequent degradation.
Molecular glues achieve the same end result—selective protein destruction—in a different way. Instead of acting as a physical bridge between the protein and the E3 ligase, molecular glues bind to one protein (often the ligase) and subtly change its shape or surface properties, improving interaction with the target protein. This induced fit causes the target protein to be ubiquitinated without a large, two-part molecule like a PROTAC.
Molecular glues are generally smaller and easier to deliver to cells, but their design is often more challenging because they rely on exploiting or discovering naturally compatible binding interfaces.
In an exciting advance in targeted protein degradation, researchers have expanded understanding of how molecular glues—particularly those derived from thalidomide—can reshape the druggable proteome. In their paper, Mining the CRBN target space redefines rules for molecular glue–induced neosubstrate recognition, Petzold et al. show that the E3 ligase substrate receptor cereblon (CRBN) can interact with more target proteins than previously imagined.
By developing bioinformatic tools to identify structural motifs compatible with CRBN recruitment, the authors predict that over 1,600 human proteins—many considered “undruggable”—could be manipulated using cereblon-binding molecular glue degraders (MGDs).
This work doesn’t just widen the therapeutic horizon; it redefines the rules of protein-ligand interaction and shows how adaptable CRBN is to diverse protein surfaces.
Using computational “matchmaking” against the structural proteome, the researchers hunted for CRBN-compatible motifs, focusing initially on the β-hairpin G-loop—a structural degron found in known CRBN neosubstrates. Their search uncovered not just hundreds of proteins with classic G-loops, but also unexpected “helical G-loops” and novel surface mimicry mechanisms that enable CRBN to bind even in the absence of these canonical features. This opens a new realm of therapeutic opportunity by expanding the CRBN target space beyond traditional degron motifs.
A Cautionary Tale: Thalidomide and it’s Dark Past
Thalidomide is an essential part of this molecular glue story. But it’s difficult to talk about thalidomide without acknowledging its dark past. In the late 1950s, thalidomide was widely distributed across Europe, Asia, Australia and parts of Africa and Latin America under various brand names, as a sedative and treatment for morning sickness. Thalidomide caused severe birth defects in thousands of children.
Thalidomide was not widely used in the U.S. The drug was submitted for FDA approval in 1960, but thanks to medical officer Dr. Frances Kelsey, the application was never approved. She demanded more safety data, particularly on neurological side effects. Because of her caution, thalidomide was never officially approved or widely distributed in the U.S. However, samples had already been given to hundreds of doctors through an “investigational” program, and it’s estimated that around 17 U.S. children were born with thalidomide-related birth defects. The legacy of thalidomide loomed over drug development for decades, serving as a chilling reminder of the need for rigorous safety testing.
And Then Redemption
In a twist of fate, thalidomide and its analogs (like lenalidomide and pomalidomide) were later found to be potent anti-cancer agents, particularly when used in multiple myeloma cases.
The key to the efficacy of thalidomide lies in its ability to bind cereblon, altering the specificity of the CRL4CRBN ubiquitin ligase complex. This results in the degradation of neosubstrate proteins that are essential for cancer cell survival. This discovery launched the field of molecular glue degraders—small molecules that facilitate proximity between a ubiquitin ligase and a target protein, triggering degradation through the proteasome.
Although thalidomide binding to cereblon was identified in 2010, the mechanism of neosubstrate recruitment—especially what makes a protein a good target for this drug—has only recently come into focus.
The paper by Petzold et al. provides the mechanistic information.
Mapping the G-Loop and Beyond: Redefining Druggable
In their report in Science, Petzold et al. took a systematic approach to discovering new CRBN-compatible proteins. Starting with the known β-hairpin G-loop motif—a loop structure with a central glycine critical for CRBN binding—they scanned both experimentally solved structures and AlphaFold2-predicted models across the human proteome. Their criteria were strict: motifs had to spatially resemble the reference loop and avoid steric clashes with CRBN.
The results were striking: They found over 6,000 G-loop-like motifs in 1,424 human proteins. Many were concentrated in zinc finger proteins, as expected, but a surprising number occurred in completely unrelated domain types—kinases, RNA-binding proteins, transcription factors, chromatin remodelers and others.
They also discovered a second structural motif they called the helical G-loop, which differs from the β-hairpin but still forms the essential hydrogen bonds with CRBN. One compelling example is the mTOR FRB domain, which contains a helical G-loop that bound CRBN in structural assays and cell-based systems.
Even more intriguing was the identification of a CRBN target that lacked any G-loop features: VAV1. This guanine nucleotide exchange factor for Rho GTPases, considered an “undruggable” protein, bound CRBN through surface mimicry of GSPT1—a well-known G- loop degron. The authors showed that the VAV1 SH3 domain engages the same key CRBN residues (N351, H357 and W400) via a completely different structural arrangement, proving that similar surface topologies, not necessarily similar folds, can enable recruitment.
A New Era of Degrader Design
By systematically validating predictions through a mix of proximity-labeling proteomics, TR- FRET, NanoBRET and X-ray crystallography, the authors confirmed many new CRBN targets, including proteins involved in splicing (HNRNPD), metabolism (ASS1), transcription (RELB) and signaling (NEK7). Interestingly, not all proteins recruited to CRBN were degraded— suggesting that engagement is necessary but not always sufficient. This opens the door to optimizing degrader chemistry to convert weak binders into effective degraders.
This study also uncovers the adaptability of the CRBN binding site itself. For instance, when recruiting NEK7, the β-hairpin G-loop is slightly offset compared to CK1α, requiring a different compound to induce degradation. VAV1, in contrast, co-opts CRBN via an entirely new interaction surface, highlighting how protein-protein interface plasticity can be exploited.
In essence, CRBN is not a rigid gatekeeper but a malleable docking hub whose interactions can be tuned via the chemical properties of MGDs and the shape of the protein surface. By embracing this flexibility, the next generation of molecular glues may be able to degrade a vast swath of the proteome once thought out of reach.
How NanoBRET Was Used in This Study
The researchers used NanoBRET (NanoLuc® Bioluminescence Resonance Energy Transfer) to validate whether their predicted protein targets formed ternary complexes with CRBN in live cells. This assay enabled them to detect real-time interactions between CRBN (fused to a HaloTag) and various target proteins (fused to NanoLuc) in the presence of different molecular glue compounds.
By measuring the energy transfer between the donor (NanoLuc) and the acceptor (HaloTag® ligand), they could confirm compound-dependent recruitment of dozens of predicted neosubstrates—including previously uncharacterized ones like NEK7, PPIL4, MNAT1, CHD7 and VAV1. Loss of the NanoBRET® signal upon mutating the G-loop glycine (or mimicked interaction site) provided strong evidence that binding was mediated through the identified structural motifs.
Conclusion
The implications of this work are in direct contrast the toxic history of thalidomide. By demonstrating that CRBN can engage targets via multiple structural modes—including canonical loops, helices and even non-loop surface mimicry—this study unlocks a rich vein of therapeutic possibilities. It’s a compelling case for how structural biology, computational modeling and chemical ingenuity can come together to turn past liabilities into future cures.
While thalidomide has a dark past, this research is redemptive—a once-feared molecule gives rise to a new era of precision degraders that may tame the most elusive of targets.
Understanding NeoSubstrate Specificity: A Key to Safe Molecular Glue Design
As the field of molecular glue degraders (MGDs) accelerates, a critical challenge remains at the forefront: understanding and controlling neosubstrate specificity. While the promise of expanding the druggable proteome is profound, the very feature that makes MGDs so powerful—their ability to reprogram E3 ligases like CRBN to degrade new proteins—can also pose significant safety risks if not carefully profiled. Off-target degradation of essential proteins, particularly in non-disease tissues, could result in unanticipated toxicity.
Among the most commonly affected neosubstrates are GSPT1, CK1α, IKZF1, IKZF2, IKZF3, and SALL4—proteins whose degradation has been linked to hematological toxicities (e.g., neutropenia and thrombocytopenia) and teratogenic effects. Identifying whether new compounds trigger degradation of these targets is a vital step in de-risking molecular glue candidates for clinical development.
To address these concerns, we offer a neosubstrate profiling platform built on HiBiT knock-in cell lines specific to these six high-risk targets. This live-cell assay system enables direct, quantitative measurement of target degradation, allowing researchers to evaluate off-target liabilities early in the discovery process. With this profiling in hand, teams can make informed decisions to optimize selectivity and confidently advance safer, more effective degraders.
Here you can explore our neosubstrate profiling services.
Kari Kenefick
