Molecular Glue Degraders
What Makes Molecular Glues Different?
Molecular glues are small, monovalent compounds that bind to an E3 ubiquitin ligase and remodel its surface to create a neo-substrate binding pocket — an entirely new protein-protein interface that does not exist without the drug. Unlike PROTACs, molecular glues have no linker and no separate target-binding warhead. The drug itself IS the glue that holds the E3 ligase and neo-substrate together.
Both major classes were discovered serendipitously: thalidomide was introduced as a sedative in the 1950s, and indisulam as a cell-cycle modulator in the 1990s. CRBN was identified as the thalidomide target in 2010 (Ito), but the neo-substrate model — IMiDs as gain-of-function ligands that recruit non-native substrates — was established in 2014 by Krönke and Lu (IKZF1/3 degradation) and Fischer (DDB1-CRBN-lenalidomide structure). DCAF15 was identified as the indisulam target in 2017. Next-generation CELMoDs such as CC-92480 (mezigdomide) are now rationally designed — using structural knowledge of the ternary interface to engineer neo-substrate selectivity and cooperativity from the outset rather than discovering it by accident. Mayor-Ruiz (2020) extended rational discovery to glue identification itself, using hyponeddylated-cell viability screens to find new glue chemotypes.
vs. PROTACs — a continuum, not a binary
- -Canonical PROTACs are heterobifunctional: two warheads on a linker, each independently engaging a pre-existing pocket
- -Canonical glues are monovalent: a single small molecule that remodels the E3 surface to recruit an otherwise unrecognized protein
- -In practice the boundary is blurred — CR8 (Slabicki 2020) is a CDK inhibitor whose solvent-exposed pyridyl glues CDK12-cyclin K directly to DDB1, and many cooperative PROTAC ternary complexes are dominated by induced E3-target PPIs rather than independent warhead binding
Key Advantages
- +Small, drug-like molecules (MW 250-600 Da) with superior oral bioavailability
- +Greatly attenuated hook effect compared to PROTACs (no competing binary warhead), though bell-shaped curves can still appear at very high glue concentrations
- +Access to "undruggable" targets that lack small-molecule binding pockets
The Neo-Substrate Mechanism
The glue molecule contributes a relatively small fraction of the total binding interface — most contacts in the ternary complex are direct protein-protein interactions between the E3 ligase and neo-substrate. This makes SAR counterintuitive: small chemical changes to the glue can completely switch neo-substrate selectivity.
No measurable affinity — surfaces are not complementary
Remodeled surface creates new PPI
CELMoDs — Cereblon E3 Ligase Modulatory Drugs
CELMoDs are the best-characterized class of molecular glue degraders. They are thalidomide analogs that bind cereblon (CRBN), a substrate receptor of the CRL4CRBN E3 ubiquitin ligase, and remodel its surface to recruit specific neo-substrate proteins for ubiquitination and proteasomal degradation.
The IMiD / CELMoD Family
All CELMoDs share a glutarimide core that anchors into CRBN's tri-tryptophan cage. Structural elaboration around this core determines which neo-substrates are recruited and with what efficiency.
| Compound | CRBN KD | Primary Neo-substrates | Clinical Status |
|---|---|---|---|
| Thalidomide | ~0.1–10 μM (assay-dependent) | SALL4 | Approved (myeloma) |
| Lenalidomide | ~0.15–5 μM (ITC ~150 nM; SPR ~1–5 μM) | IKZF1, IKZF3, CK1α | Approved (myeloma, MDS) |
| Pomalidomide | ~0.15–2 μM (ITC ~150–250 nM; SPR higher) | IKZF1, IKZF3 | Approved (myeloma) |
| Iberdomide (CC-220) | 10-100 nM (assay-dependent) | IKZF1, IKZF3 | Phase III (myeloma, SLE)* |
| Mezigdomide (CC-92480) | 5-50 nM | IKZF1, IKZF3 | Phase III (myeloma)* |
Binary KD values for CELMoD → DDB1-CRBN complex; ranges reflect literature spread across ITC and SPR/steady-state formats (Fischer 2014; Chamberlain 2014; Hartmann 2014). Note the ~100-fold improvement from lenalidomide to next-generation CELMoDs. *Clinical statuses as of 2024-2025.
Mechanism: Surface Remodeling
The glutarimide ring docks into CRBN's thalidomide-binding pocket (a hydrophobic cage formed by three tryptophan residues — Trp380, Trp386, Trp400 — plus Phe402, which lines the floor of the pocket; human CRBN numbering). The phthalimide or isoindolinone ring extends outward, presenting a modified surface. Neo-substrates engage this drug-modified surface via a small structural “degron” — typically a glycine-containing β-hairpin loop. The motif was first characterized in C2H2 zinc-finger domains (IKZF1/3, SALL4, ZFP91) but analogous structural degrons are also presented by the CK1α kinase domain and the GSPT1 (eRF3) translation termination factor.
Subtle changes — even a single amino group (lenalidomide vs. pomalidomide) — dramatically alter neo-substrate selectivity. Lenalidomide efficiently degrades CK1α while pomalidomide does not, despite differing by only one NH2 group. This underscores why SPR ternary complex profiling across panels of neo-substrates is essential for characterizing molecular glues.
Key Insight: For CELMoDs, binary CRBN affinity alone cannot predict degradation potency. Iberdomide binds CRBN ~50-fold tighter than lenalidomide, but the critical parameter is the ternary complex cooperativity factor (α) — how much the glue enhances neo-substrate recruitment. Most CELMoDs achieve α ≈ 1–20 in biophysical assays, with higher values for optimized compounds.
DCAF15 Molecular Glues — Aryl Sulfonamides
Indisulam, E7820, and tasisulam are aryl sulfonamide drugs that act as molecular glues through a different E3 ligase — DCAF15, a substrate receptor of the CRL4DCAF15 complex. They recruit the RNA splicing factor RBM39 as a neo-substrate for ubiquitination and degradation.
Indisulam: Mechanism of Action
Indisulam binds to DCAF15 (in complex with DDB1 and DDA1) and creates a composite surface that recognizes the RRM2 domain of RBM39. The drug-modified DCAF15 surface makes direct contacts with an alpha-helix in RBM39's RRM domain — contacts that are completely absent without the drug.
RBM39 degradation disrupts RNA splicing, causing cancer cell death. Notably, the closely related protein RBM23 is also degraded (sharing the RRM alpha-helix), providing a rare example of predictable neo-substrate selectivity based on structural homology.
DCAF15 vs. CRBN Glues
| Feature | CRBN (CELMoDs) | DCAF15 (Indisulam) |
|---|---|---|
| Drug scaffold | Glutarimide-based (thalidomide analogs) | Aryl sulfonamides |
| E3 complex | CRL4CRBN (DDB1-CRBN) | CRL4DCAF15 (DDB1-DCAF15-DDA1) |
| Neo-substrates | Zinc-finger proteins (IKZF1/3, SALL4, ZFP91), plus CK1α (serine/threonine kinase) and GSPT1 (eRF3 translation factor); selective GSPT1 CELMoDs include the tool compound CC-885 (Matyskiela 2016) and the clinical candidate CC-90009 / eragidomide (Surka 2021, AML trials) | RRM-domain proteins (RBM39, RBM23) |
| Degron motif | Glycine-containing β-hairpin or structural loop; characterized in C2H2 ZnF domains (IKZF1/3, SALL4) but analogous degrons also in the CK1α kinase domain and GSPT1 N-terminal domain | alpha-helix in RRM domain |
| SPR complexity | 2-protein E3 complex (DDB1-CRBN) | 3-protein E3 complex (DDB1-DCAF15-DDA1, ~200 kDa) |
CDK12-Cyclin K Glues — Direct Gluing to DDB1
A third paradigm bypasses the substrate receptor entirely: small molecules that glue CDK12-cyclin K directly to the CRL4 scaffold protein DDB1, recruiting cyclin K for ubiquitination without any DCAF.
CR8 and the “inhibitor-to-glue” transition
CR8 (Slabicki 2020, Nature 585:293) is a CDK inhibitor whose solvent-exposed 2-pyridyl moiety contacts DDB1 in the ternary complex (PDB 6TD3). It illustrates the canonical PROTAC-to-glue continuum: a monovalent kinase warhead that, by virtue of its exit-vector chemistry, also functions as a glue. Mayor-Ruiz (2020, Nat Chem Biol 16:1199) independently identified related cyclin K glues using the first rational glue-discovery platform — a differential-viability screen in hyponeddylated (CRL-inactive) cells vs WT cells, which flags compounds whose cytotoxicity depends on cullin RING ligase activity.
These two papers reframed glue discovery: no DCAF or ZnF degron is required, and glue activity can emerge from common kinase chemotypes. For SPR characterization, the practical implication is that DDB1 itself can be the immobilized E3 partner in ternary assays, with CDK12-cyclin K (or cyclin K alone) as the neo-substrate.
SPR/BLI Characterization of Molecular Glues
Molecular glues require a distinct characterization strategy compared to PROTACs. The key challenge is that the binary glue-E3 affinity is often weak (high μM for first-generation compounds), while the ternary complex is the functionally relevant species. Three assay tiers are needed.
1. Binary Binding: Glue → E3 Ligase
Measures the intrinsic affinity of the molecular glue for its E3 ligase target, independent of neo-substrate. For CELMoDs, this means measuring drug binding to the DDB1-CRBN complex.
| Parameter | Recommendation |
|---|---|
| Surface | Biotinylated DDB1-CRBN heterodimer on SA chip. Use the complex — CRBN alone is unstable and gives poor surface activity |
| Analyte | CELMoD compound (MW 250-600 Da) at 10 nM-100 μM |
| Buffer | HBS-P+ pH 7.4, 0.5-1% DMSO with solvent correction, 10 μM ZnCl2 (maintains CRBN structural integrity) |
| Density | 500-2000 RU (high density needed — small molecule analyte gives low response) |
| Analysis | Steady-state affinity for fast binders (kd > 0.1 s-1); kinetic fitting for next-gen CELMoDs with slower off-rates |
2. Ternary Complex Formation
The critical measurement for molecular glues. Three complementary approaches:
Method A: Sequential Injection
Immobilize DDB1-CRBN on SA surface. Inject CELMoD at saturating concentration (≥10× KD) and maintain in running buffer. Inject neo-substrate (e.g., IKZF1 ZnF2 domain) over the CELMoD-saturated surface. Neo-substrate binding = ternary complex formation.
Method B: Pre-incubation
Immobilize neo-substrate on surface. Pre-incubate DDB1-CRBN with CELMoD (30 min, RT), then flow the mixture over the neo-substrate surface. Compare binding signal ± CELMoD. Enhanced signal = glue-induced ternary complex. The no-glue control should give zero (or background-level) signal for a true molecular glue interaction — any baseline signal indicates a binary contaminant interaction between the E3 and neo-substrate that should be investigated.
Method C: Cooperative Binding Constant
Immobilize DDB1-CRBN. Measure neo-substrate binding at varying CELMoD concentrations (added to both analyte and running buffer). Plot KD(apparent) of neo-substrate as a function of [CELMoD]. Yields both the cooperativity factor and ternary KD.
3. Neo-Substrate Recruitment & Cooperativity (α)
The cooperativity factor is the gold standard metric for molecular glue characterization. It quantifies how much the glue enhances neo-substrate recruitment beyond what binary affinities alone would predict.
Cooperativity Factor
KDbinary = affinity of the neo-substrate for the glue-loaded E3 in the absence of cooperative contacts (often undetectable); KDternary = apparent affinity of the neo-substrate for the glue-saturated E3. The ratio captures how much the glue enhances neo-substrate recruitment beyond binary affinities alone (Douglass et al.; Hughes & Ciulli 2017).
Glue enhances neo-substrate recruitment. Most measured CELMoD cooperativities are α ≈ 1–20 in biophysical assays; values above ~20 are occasionally observed for optimized compounds (Hughes & Ciulli 2017; Kozicka 2023).
No cooperativity — compound is not functioning as a molecular glue.
Neo-Substrate Selectivity Profiling
Run a panel of neo-substrates (IKZF1-ZnF2, IKZF3-ZnF2, CK1α, SALL4-ZnF2, ZFP91-ZnF) against DDB1-CRBN with a single CELMoD. Compare ternary complex kinetics (ka, kd, KD, residence time τ = 1/kd) across neo-substrates. Present results as a heatmap of ternary KD values to visualize selectivity profiles and inform therapeutic index.
Expected Kinetics
| Interaction | KD | ka (1/Ms) | kd (1/s) |
|---|---|---|---|
| Lenalidomide → DDB1-CRBN | 2-5 μM | ~10³ | 10-2-10-1 |
| Pomalidomide → DDB1-CRBN | 0.5-2 μM | ~10³ | ~10-2 |
| Iberdomide → DDB1-CRBN | 10-100 nM | 104-105 | ~10-3 |
| CRBN-lenalidomide → IKZF1-ZnF2 | 1-20 μM | 10³-104 | ~10-1 |
| CRBN-pomalidomide → IKZF1-ZnF2 | 0.5-5 μM | 10³-104 | 10-1-10-2 |
Kinetic data compiled from Sievers et al. (2018), Hansen et al. (2020), and Matyskiela et al. (2016). Rows labeled “CRBN-[glue] → IKZF1-ZnF2” report ternary KD values — the affinity of the neo-substrate for the glue-saturated DDB1-CRBN complex. Binary affinity of IKZF1-ZnF2 for unliganded CRBN is typically undetectable. The slow apparent ka (~10³ M-1s-1) for first-generation CELMoDs reflects conformational rearrangement of CRBN's sensor loop upon glutarimide binding rather than diffusion-limited association. DCAF15 binary affinity for aryl sulfonamides is too weak to measure directly by SPR; ternary KD (DCAF15·indisulam→RBM39) is in the low-μM range (Faust 2020).
Experimental Design Considerations
Concentration Series Design
Unlike PROTACs, molecular glues do not exhibit a classical hook effect in ternary complex assays because there is no competing binary target interaction. However, at very high concentrations, free CELMoD can saturate both CRBN and any weak direct neo-substrate interaction, producing a bell-shaped curve in some configurations.
- →Binary assays: Concentration series spanning 0.1× to 100× expected KD. For lenalidomide (KD ~3 μM), use 0.3-300 μM.
- →Ternary assays: When using Method C (cooperative binding), titrate CELMoD from sub-KD to saturating. Use 8-10 concentrations for reliable cooperativity factor determination.
- →DMSO matching: CELMoDs require 0.5-1% DMSO. Run 8-point DMSO calibration curves. Even 0.05% mismatch causes ~20 RU artifact on SPR.
Surface Immobilization Strategies
DDB1-CRBN Complex
Co-express and purify as a heterodimer. Biotinylate via Avi-tag on DDB1 (N-terminus preferred). Capture on SA surface at 500-2000 RU. CRBN is a zinc-binding protein — always include 10 μM ZnCl2 in running buffer. CRBN loses activity on the surface over hours; monitor by periodic reference compound injections.
Neo-Substrate Domains
Use minimal zinc finger domains: IKZF1-ZnF2 (extended construct residues 141–196, ~6 kDa; the minimal binding unit is the ~25-residue zinc-finger hairpin), CK1α (full-length, ~39 kDa). Biotinylate or His-tag for capture. For zinc finger neo-substrates, include 10–50 μM ZnCl2 — loss of zinc = loss of fold = loss of binding (10 μM is a minimum; ZnF domains such as IKZF1-ZnF2 benefit from ≥25 μM). Use freshly purified protein (zinc fingers aggregate over days at 4°C).
DCAF15 Complex
DCAF15-DDB1-DDA1 is a large ~200 kDa three-component complex. Recombinant production is challenging — all three subunits must be co-expressed for intact complex. Biotinylate DDB1 or DDA1 for oriented SA capture. The intact three-component complex is required for relevant ternary binding data.
Ternary Complex Kinetics & Residence Time
For molecular glues, the ternary complex must persist long enough for the E3 ligase to transfer ubiquitin molecules onto the neo-substrate. Two key translatable metrics:
Longer ternary residence time generally correlates with more efficient degradation. For clinical-stage CELMoDs, report τ alongside KD.
Complexes with t1/2 > 100 s are generally considered productive for ubiquitination. This is a clinically translatable metric for CELMoD optimization.
Common Pitfalls & Solutions
| Pitfall | Description | Mitigation |
|---|---|---|
| Zinc Finger Integrity | Neo-substrates like IKZF1/3 and SALL4 bind via marginally stable zinc finger domains that lose fold without Zn2+. | Include 10 μM ZnCl2 in all buffers. Verify fold by CD or thermal shift before SPR. Use fresh protein. |
| DMSO Artifacts | CELMoDs are hydrophobic. Even 0.05% DMSO mismatch causes ~20 RU bulk refractive index shift. | 8-point DMSO calibration and solvent correction. Match DMSO exactly between sample and reference. |
| CRBN Inactivation | DDB1-CRBN loses activity on the surface over hours of data collection. | Monitor with periodic reference compound injections. Use single-cycle kinetics or capture-based approaches for long experiments. |
| Binary ≠ Cellular | Moderate binary CRBN affinity (μM) but potent cellular degradation (nM EC50). Degradation is catalytic and cooperativity amplifies effective affinity. | Always report both binary and ternary KD. The cooperativity factor α bridges the disconnect. |
| Hook Effect | At very high CELMoD concentrations, free drug in solution can compete with surface-bound complex for neo-substrate. | Include concentrations well above and below KD. If bell-shaped curve appears, focus analysis on the ascending portion. |
Key References
Structural basis of CRBN neo-substrate recruitment: Petzold G et al. (2016). “Structural basis of lenalidomide-induced CK1α degradation by the CRL4CRBN ubiquitin ligase.” Nature 532:127-130.
CELMoD mechanism & IKZF degradation: Sievers QL et al. (2018). “Defining the human C2H2 zinc finger degrome targeted by thalidomide analogs through CRBN.” Science 362:eaat0572.
DCAF15 / Indisulam: Han T et al. (2017). “Anticancer sulfonamides target splicing by inducing RBM39 degradation via recruitment to DCAF15.” Science 356:eaal3755.
DCAF15 ternary complex structure: Faust TB et al. (2020). “Structural complementarity facilitates E7820-mediated degradation of RBM39 by DCAF15.” Nat Chem Biol 16:7-14.
Next-gen CELMoD kinetics: Matyskiela ME et al. (2016). “A novel cereblon modulator recruits GSPT1 to the CRL4CRBN ubiquitin ligase.” Nature 535:252-257.
Molecular glue taxonomy: Kozicka Z et al. (2023). “The structural basis of molecular glue-induced protein degradation.” Nat Struct Mol Biol.
CRBN identified as thalidomide target: Ito T et al. (2010). “Identification of a primary target of thalidomide teratogenicity.” Science 327:1345-1350.
Neo-substrate model (IKZF1/3 degradation): Krönke J et al. (2014). “Lenalidomide causes selective degradation of IKZF1 and IKZF3 in multiple myeloma cells.” Science 343:301-305; Lu G et al. (2014). “The myeloma drug lenalidomide promotes the cereblon- dependent destruction of Ikaros proteins.” Science 343:305-309.
Structural basis of IMiD binding: Fischer ES et al. (2014). “Structure of the DDB1-CRBN E3 ubiquitin ligase in complex with thalidomide.” Nature 512:49-53.
CDK12-cyclin K glue (CR8): Slabicki M et al. (2020). “The CDK inhibitor CR8 acts as a molecular glue degrader that depletes cyclin K.” Nature 585:293-297.
Rational glue discovery platform: Mayor-Ruiz C et al. (2020). “Rational discovery of molecular glue degraders via scalable chemical profiling.” Nat Chem Biol 16:1199-1207.
Clinical GSPT1 CELMoD (CC-90009): Surka C et al. (2021). “CC-90009, a novel cereblon E3 ligase modulator, targets acute myeloid leukemia blasts and leukemia stem cells.” Blood 137:661-677.