Kinase-Independent Small-Molecule Inhibition of JAK-STAT Signaling

Abstract

Phenotypic cell-based screening is a powerful approach to small-molecule discovery, but a major challenge of this strategy lies in determining the intracellular target and mechanism of action (MoA) for validated hits. Here, we show that the small-molecule BRD0476, a novel suppressor of pancreatic β-cell apoptosis, inhibits interferon-gamma (IFN-γ)-induced Janus kinase 2 (JAK2) and signal transducer and activation of transcription 1 (STAT1) signaling to promote β-cell survival. However, unlike common JAK-STAT pathway inhibitors, BRD0476 inhibits JAK-STAT signaling without suppressing the kinase activity of any JAK. Rather, we identified the deubiquitinase ubiquitin-specific peptidase 9X (USP9X) as an intracellular target, using a quantitative proteomic analysis in rat β cells. RNAi-mediated and CRISPR/Cas9 knockdown mimicked the effects of BRD0476, and reverse chemical genetics using a known inhibitor of USP9X blocked JAK-STAT signaling without suppressing JAK activity. Site-directed mutagenesis of a putative ubiquitination site on JAK2 mitigated BRD0476 activity, suggesting a competition between phosphorylation and ubiquitination to explain small-molecule MoA. These results demonstrate that phenotypic screening, followed by comprehensive MoA efforts, can provide novel mechanistic insights into ostensibly well-understood cell signaling pathways. Furthermore, these results uncover USP9X as a potential target for regulating JAK2 activity in cellular inflammation.

Figure 1

BRD0476 suppresses inflammatory cytokine-induced human β-cell apoptosis. (a) Chemical structure of BRD0476. (b, c) Caspase-3 activity in dissociated human islets from two independent donor preparations, treated for 6 days with cytokines (a combination of IL-1β, IFN-γ, and TNF-α) and the indicated concentration of BRD0476. (d, e) Glucose-stimulated insulin secretion in dissociated human islets from two independent donor preparations, treated for 6 days with cytokines and the indicated concentration of BRD0476. Data represent the mean ± standard deviation from four individual islet donors; donor information is available in Figure S2. # p < 0.001 compared to no treatment, * p < 0.001, ** p < 0.0001 compared to cytokine treatment, Student’s t test.

Figure 2

BRD0476 inhibits JAK-STAT signaling without inhibiting the kinase activity of JAK. (a) Heat map representing individual members of the top gene set, as measured by gene-set enrichment analysis, modulated by cytokines and BRD0476 after 6 h treatment in INS-1E cells. (b) Enrichment plots and scores for gene sets representing IFN-γ-responsive genes and the JAK-STAT signaling pathway. (c) Quantitative PCR of IFN-γ-responsive genes after treatment with cytokines and 10 μM BRD0476 for the indicated times in INS-1E cells. * p < 0.05, ** p < 0.01, *** p < 0.001 versus treatment, ANOVA with Tukey corrected t test. (d) Genes present in the leading edge of >25% enriched gene sets. (e) Gamma-activated sequence-driven luciferase activity in INS-1E cells treated for 18 h with cytokines and 10 μM BRD0476. * p < 0.0001 compared to cytokine treatment with DMSO, Student’s t test. (f) Phosphorylation of STAT1 and total STAT1 protein, as measured by Western blot, in INS-1E treated for the indicated times with cytokines and 10 μM BRD0476. Tubulin was included as a loading control. (g) Biochemical kinase activity of JAK2 in the presence of the indicated concentrations of BRD0476, WP1130, or ruxolitinib.

Figure 3

BRD0476 interacts with USP9X in both β cells and cancer cells. (a) Log₂ ratios of changes of protein abundance for cells incubated with immobilized compound in the absence or presence of soluble competitor. Proteins in INS-1E cells were metabolically labeled with light and heavy amino acids lysine and arginine using SILAC methodology. Cell lysates were incubated either with BRD0476-loaded beads and 30× soluble BRD0476 or BRD0476-beads alone. Each dot represents a distinct protein. (b) Cellular ATP levels in cytokine-treated INS-1E cells following knock-down with three individual siRNA constructs for Usp9x. (c) Caspase-3 activity in cytokine-treated INS-1E cells following knock-down with three individual siRNA constructs for Usp9x. (d) Immunoblot for USP9X and actin for INS-1E wild-type and CRISPR/Cas9-knockout clone. (e) Caspase-3 activity in INS-1E wild type (black bars) and CRISPR/Cas9-knockout clone (white bars) treated for 1 day with cytokines and BRD0476. (f) Cellular ATP levels in DLD-1 colon carcinoma cells treated for 48 h with the indicated concentration of the BCL2 inhibitor ABT-737 and 10 μM BRD0476. * p < 0.0001 compared to DMSO treatment, Student’s t test.

Figure 4

USP9X interacts with JAK2 to promote cytokine signaling, which is inhibited in a kinase-independent manner by BRD0476. (a) Affinity pull-down of USP9X and associated proteins by biotinylated BRD0476. (b) Co-immunoprecipitation of USP9X and JAK2 in INS-1E cells transfected with empty vector or JAK2-FLAG. Immunoblots of input material for total USP9X and FLAG are included. (c) Phosphorylation of STAT1 and total STAT1 protein in INS-1E cells treated 30 min with cytokines and the indicated compound. Actin was included as a loading control. (d) Network graph indicating proteins connecting JAK2 to USP9X. Weight of edges is correlated with confidence level. (e) Phosphorylation of JAK2 in INS-1E cells treated with the cytokine cocktail and 10 μM BRD0476. Data represent the mean ± standard deviation. * p < 0.0001 versus control treatment, # p < 0.001 versus cytokine treatment, Student’s t test. (f) Structural analysis of JAK2 (pdb: 4GL9), with the proximity of Lys999 (blue) and Tyr1007 (red) indicated. (g) Phosphorylation of STAT1 in INS-1E cells transfected with wild-type JAK2 or K999R, followed by 30 min treatment with the cytokine cocktail and 10 μM BRD0476.