Molecular glues that inhibit deubiquitylase activity and inflammatory signalling

Abstract

Deubiquitylases (DUBs) are crucial in cell signalling and are often regulated by interactions within protein complexes. The BRCC36 isopeptidase complex (BRISC) regulates inflammatory signalling by cleaving K63-linked polyubiquitin chains on Type I interferon receptors (IFNAR1). As a Zn²⁺-dependent JAMM/MPN DUB, BRCC36 is challenging to target with selective inhibitors. We discovered first-in-class inhibitors, termed BRISC molecular glues (BLUEs), which stabilise a 16-subunit BRISC dimer in an autoinhibited conformation, blocking active sites and interactions with the targeting subunit SHMT2. This unique mode of action results in selective inhibition of BRISC over related complexes with the same catalytic subunit, splice variants and other JAMM/MPN DUBs. BLUE treatment reduced interferon-stimulated gene expression in cells containing wild type BRISC, and this effect was absent when using structure-guided, inhibitor-resistant BRISC mutants. Additionally, BLUEs increase IFNAR1 ubiquitylation and decrease IFNAR1 surface levels, offering a potential new strategy to mitigate Type I interferon-mediated diseases. Our approach also provides a template for designing selective inhibitors of large protein complexes by promoting, rather than blocking, protein-protein interactions.

Extended Data Figure 1.. Validation of hit compounds

a, Dose response curves for hit compounds against BRISC (1 nM), USP2 (100 nM) and trypsin (125 nM) using the internally-quenched fluorescence di-ubiquitin assay described in Fig. 1a. Data points from two experimental replicates are plotted. b, Re-testing of purchased H20 hit compound presumed to be AT7519. Data points are mean ± SEM from three independent experiments. c, UV-vis profile of compound in well H20 and purchased AT7519 compounds from Synkinase and Selleckchem. d, Liquid-chromatography mass spectrometry (LC-MS) spectra of H20 compound. Inset, AT7519 structure. The difference between the H20 compound and AT7519 is 173 Da, which corresponds to the mass of a dichlorobenzaldehyde group. e, MS fragmentation analyses for H20 compound and a synthesised isomer, AP-5-144. f, Profiling JMS-175-2 activity (5 μM) against a panel of 48 available DUBs using a ubiquitin-rhodamine(110)-glycine enzymatic assay.

Extended Data Figure 2.. Identification of a higher-order BRISC conformation

a, 32-point dose-response inhibition assay with JMS-175-2 and FX-171-C, with a biphasic curve fitted. Data points are mean ± SEM of two independent experiments carried out in technical duplicate. b, Michaelis-Menten plots of BRISC activity against a K63-linked di-ubiquitin fluorogenic substrate with increasing concentrations of JMS-175-2 and FX-171-C. Data points are mean ± SEM of three independent experiments carried out in technical duplicate. c, In-gel DUB assay comparing cleavage of a TAMRA-labelled K63-linked tetraubiquitin substrate by BRISC (left) and ARISC (right) with indicated compounds. For gel source data, see Supplementary Figure 1. d, Representative micrograph (BRISC dataset) and corresponding 2D class averages generated in cryoSPARC. e-f, Cryo-EM processing workflow for BRISC f, monomer and g, dimer. Green indicates selected classes for 3D refinement in cryoSPARC. f, Final monomer cryo-EM density map coloured by local resolution and Euler angular distribution (left). Rod heights are proportional to the number of particles in each direction. Unmasked FSC curves with resolution calculated using the gold standard FSC cut-off at 0.143 and 0.5 frequency. g, Final dimer maps with C1 and C2 symmetry applied, coloured by local resolution. Euler angular distribution shown with rods, and unmasked FSC curves, as in f. h, Top, surface model of ARISC dimers observed in negative stain EM from grids prepared using the GraFix cross-linking method. The same conformation is reported for BRISC dimers from nsEM grids prepared using GraFix. Bottom, an asymmetric BRISC dimer conformation observed in cryo-EM without cross-linking.

Extended Data Figure 3.. Identification of BRISC dimers in mass photometry and native mass spectrometry

a, Native mass spectra of BRISC mixed with DMSO (control), JMS-175-2, or FX-171-C. BRISC complexes and subcomplexes are highlighted. b, Table of calculated masses for different BRISC subcomplexes and super complexes. c, Mass photometry measurements of BRISC dimer at increasing inhibitor concentrations. Counts corresponding to BRISC dimer as a fraction of total counts are plotted. Data points are mean ± SEM from three independent experiments. d, Left, K63-linked diUb (dark grey) modelled on the MPN⁺ domain of BRCC36 in BRISC (light grey), based on the AMSH LP-diUb structure (PDB: 2ZNV). Right, Upon dimer formation, the second BRISC monomer sterically clashes with the proximal ubiquitin when it is bound to either BRCC36 active site.

Extended Data Figure 4.. Cryo-EM processing of the BRISC-inhibitor co-complex

Figures a-c correspond to the BRISC-FX-171-C cryo-EM dataset. Figures d-f correspond to the BRISC-JMS-175-2 dataset. a, d, Representative micrographs and 2D class averages. b, e, Image processing workflow. Green maps indicate selected classes used for 3D refinement. c, f, Left, cryo-EM density maps after 3D refinement for the final reconstructions used for model building. Asterisks indicate BLUE compound binding sites. Right, final maps with corresponding Euler angular distribution with rod heights proportional to the number of particles in each direction. FSC curves with resolution calculated using the gold standard FSC cut-off at 0.143 frequency. g, Mask used for focused refinement of the BRISC-FX-171-C map. h, Chemical structure of FX-171-C fitted into EM density before (left) and after (right) focused refinement. Cryo-EM density visualised using the surface zone tool in ChimeraX; left, radius 2.04, right, radius 2.60. i, Mask applied during refinement of BRISC-JMS-175-2 map. j, Chemical structure of JMS-175-2 fitted into EM density before (left) and after (right) focused refinement. Cryo-EM density visualised using the surface zone tool in ChimeraX; left, radius 2.20, right, radius 2.41. k, Overlay of two BRISC dimers aligned on one BRISC molecule (grey) for comparison. Models are represented as surfaces. Orange, model fitted to the BRISC-FX-171-C structure shown in c, yellow, BRISC models rigid-body fitted in the cryo-EM density of the asymmetric dimer shown in Extended Data Fig. 2g. The yellow molecule is shifted relative to the orange molecule. l, Models described in k, focussed on the small molecule binding site highlighting the shift in the BRCC45’ α6 and α10 helices.

Extended Data Figure 5.. Observed changes in BRISC subunit solvent accessibility and secondary structure in the presence of FX-171-C by HDX-MS

a, Wood’s plots generated with Deuteros showing the differences in deuterium uptake over all four HDX timepoints from three technical replicates, comparing BRISC in the absence and presence of FX-171-C. Regions highlighted in grey indicate peptides with no significant change, calculated using a 99% confidence interval, between the two conditions. The dashed line indicates the 99% confidence limit. Peptides are coloured in red to indicate deprotection in the presence of inhibitor, and blue to indicate protection. b, Peptides mapped onto BRISC dimer structure, highlighting peptides near BLUE binding site and at the interface of two BRCC45 subunits. B36 = BRCC36; B45 = BRCC45. c, Example deuterium uptake curves in the absence and presence of FX-171-C. Data points are mean ± SEM from three technical replicates. d, BLUE compounds are allosteric inhibitors and do not disrupt the BRCC36 Zn²⁺ binding site. e, CSN5 active site in complex with inhibitor CSN5i-3 (PDB: 5J0G). f, Left, BRCC45 UEV-M bound to FX-171-C aligned to Ubc13 in complex with BAY 11-7082 (PDB: 4ONN) and NSC697923 (PDB: 4ONM). Right, BRCC45 UEV-M bound to FX-171-C aligned to Cdc34 in complex with CC0651 (PDB: 3RZ3).

Extended Data Figure 6.. BLUE compounds are allosteric inhibitors and selective for human BRISC

a, Chemical structure of JMS-175-2 and analogues FX-25-1, FX-25-2, which have substitutions in the piperidine ring (highlighted in green). b, Dose-response inhibition of BRISC by indicated compounds. IC₅₀ values: JMS-175-2 = 3.8 μM, FX-25-1 = 5.2 μM, FX-25-2 = 21 μM. Data points are mean ± SEM of three independent experiments carried out in technical duplicate. c, Multiple sequence alignment (black = conserved, white = not conserved) of Abraxas1 and Abraxas2 from indicated species. Coloured boxes indicate BLUE interacting residues. d, FX-171-C inhibition of different BRISC orthologues. Hs - H. sapiens, Mm - M. musculus, Dr - D. rerio, Cf - C. floridanus. Data points are mean ± SEM of three independent experiments carried out in technical duplicate. e, f, Multiple sequence alignment of e, BRCC36 and f, BRCC45 from indicated BRISC orthologues Residues are colored as in c,. g, Mass photometry analyses of dimer formation with FX-171-C for HsBRISCΔNΔC and HsBRISCΔLoop. Fraction of counts corresponding to BRISC dimer are plotted. Data points are mean ± SEM of three independent experiments. h, Negative stain EM 2D class averages of HsBRISCΔLoop incubated with FX-171-C. 22% of particles in the 2D class averages correspond to BRISC dimers. i, BRISC-FX-171-C cryo-EM density map highlighting an extended loop in BRCC36 (dust cleaning size 7.1, map threshold 0.0044).

Extended Data Figure 7.. Determining the DUB activity, inhibitor sensitivity, and SHMT2 inhibition of structure-guided mutants

a, SDS-PAGE analysis of purified BRISC mutants. For gel source data, see Supplementary Figure 1. b, Activity of BRISC mutants against an IQF di-ubiquitin substrate. Data points are mean ± SEM of three independent experiments carried out in technical duplicate. c, FX-171-C IC₅₀ values from inhibition assays shown in Fig. 3. d, Protected and deprotected peptides from HDX-MS mapped onto the FX-171-C binding site. Peptides are coloured blue to indicate protection and red to indicate deprotection, after incubation with FX-171-C. e, Superimposition of the SHMT2 dimer from BRISC-SHMT2 structure (PDB: 6R8F) onto BRISC-FX-171-C dimer structure. SHMT2 α6 helix clashes with the BLUE binding site. f, Mutated residues in BRCC45 are not in close proximity to the SHMT2 binding site in the BRISC-SHMT2 structure (PDB: 6R8F). g, Spectral Shift (Dianthus) assays measure the binding of SHMT2(A285T) to labelled His-BRISC in the absence and presence of compounds. K_D is calculated by plotting the ratio of the fluorescence intensities at 650 nm and 670 nm against SHMT2 concentration, with a GraphPad Prism equation for one-site total binding. Data points are mean ± SEM of three independent experiments.

Extended Data Figure 8.. Establishing the effect of BLUE compound treatment on immune signalling pathways, IFNAR1 surface levels, and IFNAR1 ubiquitylation

a, Chemical structures of compounds AP-5-145, JMS-175-2, and FX-171-A. b, Dose-response inhibition of BRISC by indicated compounds. Data points are mean ± SEM for three independent experiments. c, Bar chart representing percentage of live cells across all conditions for ISRE expression and FACS analysis in THP-1 cells shown in Figs. 4a, h. THP-1 cells were treated with/without hIFNα2 (25 ng/mL) and either 4 μM inhibitor (JMS-175-2, FX-171-C, FX-171-A), 4 μM AP-5-145 negative control, DMSO control (0.1%), or JAK/STAT inhibitor Tofacitinib (*0.4 μM) for 16 hours. Bars represent the means from three independent experiments. d, e, Luciferase analysis of the ISRE in THP-1 supernatant after stimulation with d, polyI:C (1 μg/mL), or e, ODN 2216 (1 μM) and treatment with either 4 μM inhibitor (JMS-175-2, FX-171-C, FX-171-A), 4 μM AP-5-145 negative control, or JAK/STAT inhibitor Tofacitinib (*0.4 μM) for 16 hours. f, NF-κB pathway activity analysed by SEAP activity in THP-1 supernatants. Optical density measured at 625 nm. Data points in d-f are from three independent experiments. g, SgROSA MCF10A cells were treated with/without hIFN-Iα (50 ng/mL) and either 5 μM inhibitor (JMS-175-2, FX-171-C, FX-171-A), 5 μM negative control AP-5-145 or DMSO (0.1%) for 90 minutes. In g, m, IFNAR1 cell surface levels (%) were quantified using FACS analysis and calculated as a percentage of no IFN stimulation, and data points are from three independent experiments. h, BRCC45 protein levels in selected clones after knock out in MCF10A Cas9 cells. sgROSA was used as a CRISPR Cas9 control. i, BRCC45 expression in whole cell lysates from MCF10A Cas9 cells expressing CRISPR control (sgROSA), BRCC45 WT, and BRCC45 R137A. j, Anti-Flag co-immunoprecipitation performed in indicated MCF10A cell lines. BRISC complex subunits were detected using specific antibodies. k, MCF10A cell lines were treated with and without hIFNα2 (75 ng/mL) for 1 hour. STAT1 Tyr701 phosphorylation, BRCC45 and total protein levels (β-actin) were detected using specific antibodies. l, MCF10A Cas9 cells were treated with /without hIFNα2 for 4 hours. Expression of interferon-induced genes, ISG15, IFIT1, IFIT2, IFITM1, and CXCL10 were normalised to 18s rRNA and presented as fold change to own no IFN treated control. Data points are from three independent experiments. m, MCF10A cells (sgROSA, BRCC45 KO, BRCC45 WT and BRCC45 R137A) were treated with/without hIFN-Iα (50 ng/mL) for either 45 or 90 minutes. n, Anti-IFNAR1 immunoblots of TUBE-pulldown. Left, input samples after stimulation with IFNα2, treatment with BLUE inhibitors and cell lysis. Right, ubiquitylated IFNAR1 isolated with agarose-TUBE beads after IFNα2 stimulation. The Western blot shown is representative of three biological replicates. For raw, uncropped Western blots, see Supplementary Figure 1. o, Ponceau stained membranes of blots shown in n, prior to antibody incubation. p, Densitometry quantification of Western blot shown in n, and two other biological replicates (n=3). In d-f, paired t-tests were used to compare compound treated cells with DMSO control cells. In g, unpaired t-tests were used to compare compound treated cells with DMSO control cells. In l, a two-way ANOVA was used to gene expression levels for five genes in both BRCC45 WT and BRCC45 R137A cells to the sgROSA MCF10A cells. In m, a two-way ANOVA was performed to compare no IFN vs. IFN 45 minutes vs. IFN 90 minutes. P values illustrated by * <0.05, ** <0.01, *** <0.005, **** <0.0001, ns = non-significant. Error bars represent ± SEM.

Extended Data Figure 9.. BLUE compounds reduce interferon-stimulated gene expression in stimulated healthy and unstimulated SSc PBMCs

a, c, d, Type I IFN signalling gene expression analysis of healthy control PBMCs treated with/without IFNα2 (20 ng/mL) and DMSO control (0.1 %) or 2 μM AP-5-145, JMS-175-2, or FX-171-C for 16 hours (n=3). a, Heatmap of each ISG expression levels relative to each donors housekeeping gene expression levels (geomean of ACTB, GAPDH, HPRT1, RPLP0), shown as Log2 fold change to grouped AP-5-145. Data shown for each individual donor, HC = healthy control. Heat map represents the mean fold change from three healthy donors.c, Volcano plot illustrating genes increased with addition of IFN + DMSO vs. DMSO only. d, Volcano plot illustrating no change in gene expression with negative control AP-5-145 + IFN vs. DMSO + IFN only. In c, and d, data points are the means from three independent experiments. b, As in a, Type I IFN signalling gene expression analysis of unstimulated SSc PBMCs from nine patients, treated with 2 μM AP-5-145, JMS-175-2, or FX-171-C for 16 hours. ISG relative expression to each donors housekeeping genes, shown as Log2 fold change relative to grouped AP-5-145, as in a. P refers to patient number i.e. P1 = patient 1. Heat map represents the mean fold change from nine SSc donors. e, PBMCs were isolated from patients and treated with DMSO (0.1%), 2 μM AP-5-145, FX-171-C or JMS-175-2 for 16 hours without IFN stimulation. Composite ISG score (including CXCL10, IFIT1, ISG15 and MX1) gene expression analysis between conditions relative to each donor DMSO control. Error bars represent ± SEM. Inividual data points represent the mean fold change for each gene for 20 donors.

Extended Data Figure 10.. Proposed model of BLUE compound mode of action

Interferon binding to IFNAR1 receptors triggers JAK/STAT signalling and an elevated immune response. Interferon also initiates IFNAR1 receptor ubiquitylation (K63-linked), receptor internalisation and lysosomal degradation. The BRISC-SHMT2 complex is required for deubiquitylation of IFNAR1. BRISC is recruited to IFNAR1/2 through interactions with SHMT2 to promote sustained interferon signalling and inflammation. BLUE compounds (blue stars) promote formation of a BRISC dimer complex, which sterically hinders SHMT2 and polyubiquitin binding.

Figure 1.. Fluorescence-based screen to identify first-in-class JAMM inhibitors

a, Schematic of a TAMRA-linked internally quenched fluorescent (IQF) di-ubiquitin substrate (left) and reaction progress curve of BRISC DUB activity (right). b, Z-score normalisation of 320 compounds from an in-house kinase-directed inhibitor library and identification of hit compounds in wells H20 and P12. SD = standard deviation. c, Chemical structures of AT7519 and of two isomers with an additional 2,6-dichlorobenzaldehyde moiety. d, Dose-response inhibition of BRISC activity by the H20 compound and the two potential isomers, AP-5-144 and JMS-175-2. e, g, Dose-response inhibition of trypsin, USP2 and JAMM/MPN DUB enzymes AMSH* (a STAM2-AMSH fusion), BRISC, ARISC, and BRCC36-Abraxas2 by the indicated compounds. Data points in d, e, and g, are mean ± SEM of three independent experiments carried out in technical duplicate. f, Chemical structure of the FX-171-C compound.

Figure 2.. Inhibitors stabilise an inhibited BRISC dimer

a, Mass photometry histograms of purified BRISC in absence (DMSO, top) and presence of inhibitors (JMS-175-2, middle; FX-171-C, bottom), and corresponding negative stain EM 2D classes of BRISC mixed with DMSO or inhibitors (insets). b, Left, cryo-EM density map of a BRISC monomer with BRISC model (PDB: 6H3C) rigid-body fitted (dust cleaning size 7.4, map threshold 0.0907). Right, cryo-EM density map of a BRISC dimer with two BRISC models rigid-body fitted. Maps are outputs from non-uniform refinement in cryoSPARC. c, Cryo-EM density map of BRISC-FX-171-C co-structure at 3.0 Å. BRISC monomers are shown as grey and salmon cartoon models and fitted to the cryo-EM map shown as a transparent surface at 0.00224 threshold. The C-termini of BRCC45 (residues 275–383) and MERIT40 are rigid-body fitted into the density. d, BRISC-FX-171-C cryo-EM density map at 0.0165 threshold. BRISC subunits are coloured by chain. The density corresponding to FX-171-C is coloured orange and highlighted in orange boxes. The map shown in c, and d, is a locally-filtered map generated using RELION local resolution estimation. e, Close-up views of the indicated inhibitor density comparing FX-171-C (left) and JMS-175-2 (right) binding sites. f, Cryo-EM density at the equivalent sites of BRCC36, Abraxas2, and BRCC45’ in the BRISC-FX-171-C co-structure where there is no dimer interface, and no additional density corresponding to FX-171-C. The maps in d-f had dust cleaning (size 7.1) applied in ChimeraX. g, Structures of FX-171-C and JMS-175-2 modelled in State 1 and State 2. Cryo-EM density of the inhbitor after focused refinement represented as a mesh and displayed using the surface zone tool (FX-171-C radius 2.6, JMS-175-2 radius 2.2) in ChimeraX.

Figure 3.. Analysis of the BLUE compound binding site

a, Ball and stick model of FX-171-C and JMS-175-2 binding to BRCC36, Abraxas2 and BRCC45. Hydrogen bonds shown as black dashed lines and residues studied by mutagenesis are indicated. b, The BLUE compound binding pocket shown as a surface and coloured by hydrophobicity. c, FX-171-C inhibition of BRISC DUB activity with BRCC36, Abraxas2 and BRCC45 mutants. d, SHMT2 inhibition of the same BRISC mutants as in c,. Data in c, and d, are mean ± SEM of three independent experiments carried out in technical duplicate.

Figure 4.. BLUE compounds reduce interferon-stimulated gene (ISG) expression and IFNAR1 internalisation in cells

a, THP-1 cells were treated with/without hIFNα2 (25 ng/mL) and either 4 μM inhibitor (JMS-175-2, FX-171-C, FX-171-A), 4 μM negative control AP-5-145, DMSO control (0.1%), or JAK/STAT inhibitor Tofacitinib (*0.4 μM) for 16 hours. Luciferase analysis of the ISRE in THP-1 supernatant in relative light units (RLU). Data points are from four independent experiments. b-f, MCF10A Cas9 cells expressing BRCC45 wild-type (WT) and BRCC45 R137A were treated with hIFNα2 (75 ng/mL) and either 2.5 μM inhibitor (JMS-175-2, FX-171-C, FX-171-A), 2.5 μM negative control AP-5-145 or DMSO (0.1%) for 4 hours. Expression of indicated interferon-induced genes (b, ISG15, c, IFIT1, d, IFIT2, e, IFITM1, f, CXCL10) normalised to 18s rRNA are presented as fold change to own IFN + DMSO treated control. Data points are from four independent experiments. g, MCF10A cells (BRCC45 WT and BRCC45 R137A) were treated with/without hIFN-Iα (50 ng/mL) and either 5 μM inhibitor (JMS-175-2, FX-171-C, FX-171-A), 5 μM negative control AP-5-145 or DMSO (0.1%) for 90 minutes. IFNAR1 cell surface levels (%) were quantified using FACS analysis and calculated as a percentage of no IFN stimulation. Data points are from three independent experiments. h, THP-1 cells were treated with/without hIFNα2 (25 ng/mL) and either 4 μM inhibitor (JMS-175-2, FX-171-C, FX-171-A), 4 μM negative control AP-5-145, DMSO control (0.1%), or JAK/SAT inhibitor Tofacitinib (*0.4 μM) for 16 hours. IFNAR1 surface levels were quantified using FACS analysis and median fluorescence intensity of allophycocyanine(APC)-IFNAR1 calculated as a percentage to no IFN stimulation. Data points are from three independent experiments. Statistical analyses in a, were performed using paired t-tests to compare compound treated cells to DMSO control cells. In b-g, one-way ANOVA with Dunnett’s multiple comparisons test was used to compare statistical significance between AP-5-145 and BLUE compound treatment. In h, unpaired t-tests were used to compare compound treated cells to DMSO control cells. P values illustrated by * <0.05, ** <0.01, *** <0.005, **** <0.0001, ns = non-significant. Error bars represent ± SEM.

Figure 5.. BLUE compounds reduce interferon-stimulated gene (ISG) expression in PBMCs

a, b, Type I IFN signalling gene expression analysis (67 genes normalised for housekeeping genes: ACTB, GAPDH, HPRT1, RPLP0) of healthy control PBMCs stimulated with IFNα2. Volcano plot of genes increased with addition of IFNα2 with negative control AP-5-145 compared to DMSO (no IFN) condition, and b, effect of JMS-175-2 + IFN stimulation compared to AP-5-145 + IFN. Blue line indicates a p-value of 0.05. Data points are the means from three independent experiments. c, CXCL10 protein levels in supernatant from IFNα2-stimulated healthy PBMCs (n=3) quantified by ELISA and shown as a percentage of own IFN + DMSO control (100%). Bar graph is average of three independent experiments. d, PBMCs were isolated from SSc patients and treated with DMSO, AP-5-145, FX-171-C, or JMS-175-2 for 16 hours without IFN stimulation. Secreted CXCL10 in supernatant is shown as percentage to own DMSO control. Bar graph is average of data from seven SSc donors. e, Type I IFN signalling gene expression analysis of unstimulated SSc PBMCs, treated with 2 μM AP-5-145, JMS-175-2, or FX-171-C. Heat map showing log2 mean fold change in ISG expression by treatment, compared to AP-5-145, via qPCR SuperArray. ΔCt was calculated against the geometric mean of four housekeeping genes, followed by ΔΔCt (fold change) relative to AP-5-145, and log2 transformation. Heat maps represent the mean fold change from nine SSc donors. In a, and b, paired two-tailed student t-tests were used to compare between treatment conditions for statistical significance. P values illustrated by * <0.05, ** <0.01, ns = non-significant. Error bars represent ± SEM.