CSN5i-3 is an orthosteric molecular glue inhibitor of COP9 signalosome
Summary
This Nature paper resolves how the COP9 signalosome (CSN) recognises neddylated cullin–RING ligases (N8~CRLs) in a pre-catalytic state and reveals a surprising mechanism of inhibition by CSN5i-3. Using high-resolution cryo-EM of CSN mutants and inhibitor-bound complexes, the authors show that CSN5i-3 binds the CSN5 active site (orthosterically) yet also makes direct contacts with the NEDD8 (N8) tail and the CUL1–WHB domain. That dual action stabilises the enzyme–substrate assembly — i.e. the compound functions as a molecular glue — and explains its unusually high, substrate-dependent potency despite only moderate affinity for the free enzyme.
Article Date: 11 February 2026
Article URL: https://www.nature.com/articles/s41586-026-10129-y
Article Image: https://media.springernature.com/lw685/springer-static/image/art%3A10.1038%2Fs41586-026-10129-y/MediaObjects/41586_2026_10129_Fig1_HTML.png
Key Points
- High-resolution cryo-EM structures of CSN DM (catalytically disabled mutant) bound to multiple N8~CRLs reveal the pre-catalytic state with the N8 C-terminal tail threaded across the CSN5 catalytic cleft.
- CSN5i-3 is orthosteric: it coordinates the catalytic Zn2+ and occupies the active site, sterically clashing with the N8–cullin iso-peptide linkage.
- Despite only micromolar affinity for free CSN/CSN5–CSN6, CSN5i-3 inhibits deneddylation with low-nanomolar potency by stabilising the CSN–N8~CRL assembly — acting as an orthosteric molecular glue (OMG).
- The inhibitor simultaneously contacts the N8 C-terminal tail and the CUL1–WHB domain, extending the N8–CSN5 exosite interface and thereby increasing cooperative binding.
- In cells CSN5i-3 reshapes the CSN interactome: it increases CSN association with many CRLs and causes differential stability of substrate receptors, consistent with trapped enzyme–substrate complexes and downstream changes in substrate turnover.
- The OMG concept suggests a route to substrate-selective orthosteric inhibitors that combine active-site tractability with substrate-dependent specificity — a potentially useful strategy in drug design to reduce off-target effects.
Content summary
The authors used a catalytically dead CSN variant (E76A/D151N) to capture the pre-catalytic CSN–N8~CRL complexes at ~3.0–3.5 Å resolution, resolving the N8 tail and detailing contacts from CSN5 Ins-1/Ins-2 loops, the CUL WHB domain and the RBX RING domain. They then determined structures of CSN bound to CSN5i-3 alone and to CSN5i-3 plus N8~CRL1. CSN5i-3 occupies the catalytic cleft and buries the zinc ion, but also wedges between CSN5, the N8 tail and CUL1–WHB, stabilising the assembly that it ostensibly blocks. Biophysical assays (BLI, ITC) show micromolar binding to free enzyme but much stronger, inhibitor-dependent N8 binding; enzymatic assays show nanomolar IC50s. XL-MS and cellular AP–MS confirm the inhibitor-dependent stabilisation and altered CSN–CRL interactions in cells. The paper defines these dual properties and proposes the orthosteric molecular glue (OMG) concept.
Context and relevance
CSN regulates cullin–RING E3 ligases via deneddylation, a central node in ubiquitin-mediated proteostasis and targeted protein degradation strategies. Understanding how CSN recognises N8~CRLs at an atomic level fills a structural gap for the deneddylation mechanism and exposes exploitable exosites. The discovery that an active-site binder can act as a substrate-dependent molecular glue alters how we think about orthosteric inhibitors: instead of being strictly substrate-agnostic, they can be engineered to harness substrate contacts and achieve selectivity. That has immediate implications for chemical biology and drug discovery, especially for enzymes with adjacent exosites or modular substrates (e.g. proteases, de-modifying enzymes, E3 regulators).
Why should I read this?
Short version — if you care about enzyme inhibitors or targeted degradation, this paper is a neat bit of molecular trickery. The team reads like structural biology on steroids: resolving a stubborn pre-catalytic state and showing an inhibitor that not only blocks the active site but glues the enzyme to its substrate. It’s a clever mechanism that could be used to make more selective drugs or to predict how inhibitors will change cellular networks. Read it for the structures, the unexpected mechanism, and the drug-design idea (OMG inhibitors) that could save you time in future lead-optimisation work.
Author note
Punchy take: the authors don’t just solve a structure — they flip an inhibitor class on its head and propose a concrete, actionable concept for substrate-selective orthosteric drugs. Important for med-chem and anyone designing molecule-to-protein interfaces.