Basis for metabolite-dependent Cullin-RING ligase deneddylation by the COP9 signalosome

Abstract

The Cullin-RING ligases (CRLs) are the largest family of ubiquitin E3s activated by neddylation and regulated by the deneddylase COP9 signalosome (CSN). The inositol polyphosphate metabolites promote the formation of CRL-CSN complexes, but with unclear mechanism of action. Here, we provide structural and genetic evidence supporting inositol hexakisphosphate (IP₆) as a general CSN cofactor recruiting CRLs. We determined the crystal structure of IP₆ in complex with CSN subunit 2 (CSN2), based on which we identified the IP₆-corresponding electron density in the cryoelectron microscopy map of a CRL4A-CSN complex. IP₆ binds to a cognate pocket formed by conserved lysine residues from CSN2 and Rbx1/Roc1, thereby strengthening CRL-CSN interactions to dislodge the E2 CDC34/UBE2R from CRL and to promote CRL deneddylation. IP₆ binding-deficient Csn2^K70E/K70E knockin mice are embryonic lethal. The same mutation disabled Schizosaccharomyces pombe Csn2 from rescuing UV-hypersensitivity of csn2-null yeast. These data suggest that CRL transition from the E2-bound active state to the CSN-bound sequestered state is critically assisted by an interfacial IP₆ small molecule, whose metabolism may be coupled to CRL-CSN complex dynamics.

Keywords: COP9 signalosome; Cullin-RING ligases; deneddylation; inositol hexakisphosphate; intermolecular glue.

Fig. 1.

IP₆ is a CSN cofactor. (A–C) Quantitative ITC measurement of the interaction between IP₆ and CSN2 (A), Cul4A/Rbx1 (B), or CSN2–Cul4A/Rbx1 complex (C). n.d.: Not determined because of weak binding. (D) Mass spectra of myc-CSN2 immunoprecipitate. (Lower) Zoom-in spectra highlighting the mass corresponding to [IP₆]⁻². (Inset) CSN2 Western blot of wild-type and myc-CSN2 stable cells. (E) Deneddylation of purified Nedd8–Cul4A/Rbx1 by recombinant CSN prepared from E. coli (IP₆-free), in the presence/absence of IP₆. (F and G) In-lysate Cul4A deneddylation at the indicated time points in the absence/presence of exogenous IP₆ (F), or in control and CRISPR-mediated IP5K KO cells (G). Repeated three times. Cells were lysed in the absence of protease inhibitor to keep CSN activity. Note that some deneddylation occurs during cell lysis and before experimental time “0”. (Lower) Scatter plot quantifying the half-life of Nedd8-Cul4A. l.e., long exposure s.e., short exposure.

Fig. 2.

Structural basis of the IP₆–CSN2 complex. (A) Surface representation of the IP₆–CSN2–T4 crystal structure. IP₆ is shown as stick with cyan backbone and buried in an electropositive (blue) pocket. (B) The IP₆–CSN2 crystal structure highlighting a 2F_o − F_c omit electron density map of IP₆ contoured at 1.2σ. (C) Rotated view of the IP₆–CSN2 complex highlighting the cluster of basic residues coordinating IP₆. (D) Sequence alignment of IP₆-coordinating residues from Rbx1/2 across species. At: Arabidopsis thaliana; Ce: Caenorhabditis elegans; Dm: Drosophila melanogaster; Dr: Danio reiro; Gg: Gallus gallus; Hs: Homo sapiens; Mm: Mus musculus; Sc: Saccharomyces cerevisiae; Sp: Schizosaccharomyces pombe. (E) In vitro binding between HEK293-purified, salt-stripped GST–Cul4A/Rbx1 and bacterially purified recombinant CSN2 variants that are defective in binding IP₆. The mutants fail to show augmented Cul4A/Rbx1 binding in the presence of IP₆. (F) GST pulldown of CSN2 wild-type and IP₆ binding-defective mutants from cell lysates.

Fig. 3.

Structural and functional insights into the IP₆–CSN–CRL interface. (A) Fitting of the Nedd8–CRL4A–CSN electron density map (EMD-3314, 6.4 Å) with crystal structures of the subcomponents. (B) Close-up view of the IP₆–CRL–CSN ternary complex. IP₆ is positioned based on the IP₆–CSN2 crystal structure. (C) Electrostatic complementarity between IP₆ and its basic binding pocket. (D) Sequence alignment of IP₆-coordinating residues from Rbx1/2 across species. At: Arabidopsis thaliana; Ce: Caenorhabditis elegans; Dm: Drosophila melanogaster; Dr: Danio reiro; Gg: Gallus gallus; Hs: Homo sapiens; Mm: Mus musculus; Sc: Saccharomyces cerevisiae; Sp: Schizosaccharomyces pombe. (E) GST pulldown of Cul4A coexpressed with Rbx1 wild-type or mutants. (F) In vitro binding between recombinant CSN2 and HEK293-purified GST–Cul4A coexpressed with myc-Rbx1 wild-type or the various mutants, with/without IP₆. (G) Molecular structures of biotin (2)-IP₆ (i) and biotin (5)-IP6 (ii). For synthesis details, see extended data. (H) Biotin (5)-IP₆ but not biotin (2)-IP₆ substitutes IP₆ in stimulating in vitro interaction between purified recombinant CSN2 and Cul4A/Rbx1. (I) Pulldown of CSN2 and Cul4A from whole cell lysate by biotin (5)-IP₆ but not biotin (2)-IP₆.

Fig. 4.

IP₆ promotes CSN-CRL4 complex reassembly by augmenting competition against CDC34, thereby protecting substrate receptor. (A) Recombinant purified CDC34 dose-dependently competes with CSN2 for Cul4A/Rbx1 binding in vitro. (B) IP₆ facilitates CSN2 to compete with CDC34 for binding to Cul4A/Rbx1. (C) Relative binding of Cul4A/Rbx1 to immobilized CSN2 in the presence/absence of CDC34 and/or IP₆ (***P < 0.001, student’s t test; n.s., not significant). An anti-Cul4A antibody was used to detect the ELISA (enzyme-linked immunosorbent assay) signal. (D) IP5K knockdown diminishes the interaction between Cul4A and CSN2. (E) IP5K knockdown enhances Cul4A-CDC34 interaction while diminishes the levels of DDB2. (F) Effects of overexpressing Rbx1 wild-type or the IP₆-binding defective K25E/K26E mutant on Cul4A neddylation and DDB2 levels. (G) Scheme depicting the role of IP₆ in facilitating the sequestration of CRL by CSN, thereby completing the CRL activity cycle. Other proposed CRL activation steps would include: 1) Substrate-induced CSN-dissociation (1′: CAND1-mediated adaptor exchange) (11, 13); 2) Neddylation; 3) ARIH-mediated mono-Ub priming (50); 4) CDC34-dependent Ub-chain elongation and substrate degradation (23). N8: Nedd8; R: Rbx1/Roc1; SR: substrate receptor; Sub: substrate; Ub: ubiquitin; 2: CSN2.

Fig. 5.

IP₆ is a general CRL regulator conserved from yeast to mammals. (A–C) GST pulldown of cellular Cul1 (A), Cul2 (B), Cul3 (C) coexpressed with Rbx1 wild-type or mutants deficient in IP₆ coordination. Rbx1 mutations disrupting IP6 coordination also abolish CSN interaction with Cul1–3. (D) GST pulldown of CSN2 wild-type and IP₆ binding-defective mutants from cell lysates, CSN2 mutations disrupting IP₆ coordination also fail to pull down Cul1–5. (E) Deneddylation of purified Nedd8–Cul3/Rbx1 by recombinant CSN, with/without IP₆ at the indicated concentrations. (F) Levels of Cullin neddylation in wild-type and IP5K KO cells. *P < 0.05, **P < 0.01. (G) Coimmunoprecipitation between myc-Cul1 and TAP-Csn2, each tagged at endogenous locus in wild-type and Δipk1 S. pombe strains. Cul–Csn2 interaction is abolished in the Δipk1 mutant. (H) UV sensitivity of a Δcsn2 strain rescued with wild-type Csn2, but not the K70E mutant. **P < 0.01.