A proximity biotinylation-based approach to identify protein-E3 ligase interactions induced by PROTACs and molecular glues

Abstract

Proteolysis-targeting chimaeras (PROTACs) as well as molecular glues such as immunomodulatory drugs (IMiDs) and indisulam are drugs that induce interactions between substrate proteins and an E3 ubiquitin ligases for targeted protein degradation. Here, we develop a workflow based on proximity-dependent biotinylation by AirID to identify drug-induced neo-substrates of the E3 ligase cereblon (CRBN). Using AirID-CRBN, we detect IMiD-dependent biotinylation of CRBN neo-substrates in vitro and identify biotinylated peptides of well-known neo-substrates by mass spectrometry with high specificity and selectivity. Additional analyses reveal ZMYM2 and ZMYM2-FGFR1 fusion protein-responsible for the 8p11 syndrome involved in acute myeloid leukaemia-as CRBN neo-substrates. Furthermore, AirID-DCAF15 and AirID-CRBN biotinylate neo-substrates targeted by indisulam and PROTACs, respectively, suggesting that this approach has the potential to serve as a general strategy for characterizing drug-inducible protein-protein interactions in cells.

Fig. 1. AirID-CRBN biotinylates neo-substrates in the presence of thalidomide and IMiDs in cells.

a Chemical structures of IMiDs used in this study. b Schematic diagram of IMiDs-dependent neo-substrate biotinylation by AirID-CRBN. c IMiDs-dependent biotinylation assay of exogenous neo-substrates by AirID-CRBN in cells. HEK293T cells stably expressing AGIA-AirID-CRBN-WT or -YW/AA were transfected with pCDNA3.1-Myc-IKZF1 and -SALL4, and treated with DMSO or 10 µM pomalidomide (Po), 5 µM biotin and 10 µM MG132 for 6 h. The experiment was repeated three times independently with similar results. d LC-MS/MS analysis of biotinylated peptides or proteins using three enrichment methods. MM1.S cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO or 10 µM pomalidomide (Po) in the presence of 10 µM biotin and 5 µM MG132 for 8 h (biological replicates; n = 3). Biotinylated peptides were enriched with tamavidin 2-REV or anti-biotin antibody, and biotinylated proteins were enriched with streptavidin. Significant changes in the volcano plots were calculated by Student’s two-sided t-test and the false discovery rate (FDR)-adjusted P-values calculated using Benjamini–Hochberg method are shown in the Supplementary Data 2–4. e Neo-substrate selectivity of IMiDs for biotinylation by AirID-CRBN. MM1.S cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO, 20 µM thalidomide (Th), 10 µM lenalidomide (Le), 10 µM pomalidomide (Po) or 20 µM 5-hydroxythalidomide (5HT) in the presence of 10 µM biotin and 5 µM MG132 for 8 h. The experiment was repeated twice independently with similar results. f LC-MS/MS analysis of biotinylated peptides using AirID-CRBN. MM1.S cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO, 20 µM thalidomide (Th), 10 µM lenalidomide (Le), 10 µM pomalidomide (Po) or 20 µM 5-hydroxythalidomide (5HT) in the presence of 10 µM biotin and 5 µM MG132 for 8 h (biological replicates; n = 3). Biotinylated peptides were enriched with tamavidin 2-REV and analysed by LC-MS/MS. g Heat map of biotinylated peptides of IKZF1 or IKZF3 (IMiD/DMSO ratio >5 and P-value < 0.05) among IMiDs detected in LC-MS/MS analysis using MM1.S cells. f, g Significant changes in the volcano plots and heat map were calculated by Student’s two-sided t-test and the false discovery rate (FDR)-adjusted P-values calculated using Benjamini–Hochberg method are shown in the Supplementary Data 5. c, e Biotinylated proteins were pulled down using streptavidin beads and analysed by immunoblotting. Source data are provided as a Source data file.

Fig. 2. AirID-CRBN with thalidomide and IMiDs biotinylates neo-substrates in cells in a highly selective manner.

a LC-MS/MS analysis of biotinylated peptides using AirID-CRBN. HEK293T cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO, 20 µM thalidomide (Th), 10 µM lenalidomide (Le), 10 µM pomalidomide (Po) or 20 µM 5-hydroxythalidomide (5HT) in the presence of 10 µM biotin and 5 µM MG132 for 8 h (biological replicate; n = 3). b Heat map of biotinylated peptides (ThD/DMSO ratio >3 and P-value <0.05) in HEK293T cells detected by LC-MS/MS analysis. a, b Significant changes in the volcano plots and heat map were calculated by Student’s two-sided t-test and the false discovery rate (FDR)-adjusted P-values calculated using Benjamini–Hochberg method are shown in the Supplementary Data 6. c IMiD-dependent biotinylation assay of neo-substrate candidates detected by LC-MS/MS analysis in HEK293T cells. HEK293T cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO, 20 µM thalidomide (Th), 10 µM lenalidomide (Le), 10 µM pomalidomide (Po) or 20 µM 5-hydroxythalidomide (5HT) in the presence of 10 µM biotin and 5 µM MG132 for 8 h. Then, the biotinylated proteins were pulled down using streptavidin beads and analysed by immunoblotting. The experiment was repeated three times independently with similar results. d Immunoblot analysis of endogenous protein levels of neo-substrates in HEK293T cells treated with DMSO, 10 µM thalidomide (Th), 10 µM lenalidomide (Le), 10 µM pomalidomide (Po), 10 µM 5-hydroxythalidomide (5HT) or 1 µM CC-885 for 24 h. The experiment was repeated twice independently with similar results. Source data are provided as a Source data file.

Fig. 3. Biotin-dependent LS-MS/MS analyses using HuH7 and IMR32 cells expressing AirID-CRBN.

a, b LC-MS/MS analysis of biotinylated peptide using AirID-CRBN for (a) HuH7 or (b) IMR32 cells. HuH7 or IMR32 cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO, 20 µM thalidomide (Th), 10 µM lenalidomide (Le), 10 µM pomalidomide (Po) or 20 µM 5-hydroxythalidomide (5HT) in the presence of 10 µM biotin and 5 µM MG132 for 8 h (biological replicates; n = 3). c, d Heat map of biotinylated peptides (IMiDs/DMSO ratio >3 and P-value <0.05) in c HuH7, or d IMR32 cells detected by LC-MS/MS analysis. a–d Significant changes in the volcano plots and heat maps were calculated by Student’s two-sided t-test and the false discovery rate (FDR)-adjusted P-values calculated using Benjamini–Hochberg method are shown in the Supplementary Data 7, 8. e, f IMiDs-dependent biotinylation assay of neo-substrate candidates detected by LC-MS/MS analyses in e HuH7 or f IMR32 cells. HuH7 or IMR32 cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO, 20 µM thalidomide (Th), 10 µM lenalidomide (Le), 10 µM pomalidomide (Po) or 20 µM 5-hydroxythalidomide (5HT) in the presence of 10 µM biotin and 5 µM MG132 for 8 h. Then, the biotinylated proteins were pulled down using streptavidin beads and analysed by immunoblotting. The experiments were repeated twice independently with similar results. Source data are provided as a Source data file.

Fig. 4. AirID-CRBN with thalidomide and IMiDs presents neo-substrate candidates in cells.

a Schematic diagram of AlphaScreen-based biochemical assay for detecting IMiDs-dependent interaction between CRBN and neo-substrates. b In vitro interaction assay between CRBN and neo-substrate candidates. The interaction between biotin-labelled bls-CRBN and FLAG-GST-neo-substrate candidates in the presence of DMSO or 20 µM thalidomide (Th), lenalidomide (Le), pomalidomide (Po) or 5-hydroxythalidomide (5HT) was analysed using the AlphaScreen-based biochemical assay. All AlphaScreen signals mean raw luminescent signal in the AlphaScreen-based biochemical assay. Error bars denote the standard deviation (independent experiments; n = 3). c Immunoblot analysis of IMiD-dependent protein degradation of ZNF536, ZNF687 and ZMYM2. HEK293T cells expressing AGIA-ZNF536, Myc-ZNF687 or ZMYM2-AGIA and FLAG-CRBN were treated with DMSO, thalidomide (Th), lenalidomide (Le), pomalidomide (Po) or 5-hydroxythalidomide (5HT) for 16 h. The experiment was repeated three times independently with similar results. d Immunoblot analysis of endogenous ZMYM2 protein levels in IMR32 cells treated with DMSO, thalidomide (Th), lenalidomide (Le), pomalidomide (Po), 5-hydroxythalidomide (5HT) or CC-885 for 24 h. The experiment was repeated three times independently with similar results. e IMR32 and HEK293T cells were treated with DMSO or pomalidomide (Po) for 24 h and ZMYM2 mRNA expression levels were measured by quantitative RT-PCR. Relative mRNA expression was determined using the expression level with DMSO treatment. Error bars denote the standard deviation (biological replicates; n = 4). f Immunoblot analysis of ZMYM2 protein levels in IMR32 and HEK293T cells treated with DMSO or pomalidomide (Po) in the presence of DMSO, 5 µM MG132 or 1 µM MLN4924 for 24 h. g Immunoblot analysis of ZMYM2 protein levels in IMR32 cells treated with DMSO or pomalidomide (Po) for 48 h ZMYM2 band intensity was normalised to tubulin band intensity. All relative band intensities are expressed as fluorescent signals relative to that of DMSO. Error bars denote the standard deviation (biological replicates; n = 3) and P-values were calculated by one-way ANOVA with Tukey’s post-hoc test (*P = 0.031). Source data are provided as a Source data file.

Fig. 5. Comparison of biotinylation enzymes in biotinylation of neo-substrate.

a, b IMiDs-dependent biotinylation assay of neo-substrates by BioID-CRBN in a HEK293T or b IMR32 cells. HEK293T or IMR32 cells stably expressing AGIA-BioID-CRBN were treated with DMSO or 10 µM pomalidomide (Po) in the presence of 10 µM biotin and 5 µM MG132 for 2 h or 6 h. The experiments were repeated twice independently with similar results. c, d IMiDs-dependent biotinylation assay of neo-substrates by AirID-CRBN or TurboID-CRBN in c HEK293T or d IMR32 cells. HEK293T or IMR32 cells stably expressing AGIA-AirID-CRBN or AGIA-TurboID-CRBN were treated with DMSO or 10 µM pomalidomide (Po) in the presence of 10 µM biotin and 5 µM MG132 for 2 h or 6 h. The experiments were repeated twice independently with similar results. e Comparison between AirID-CRBN and TurboID-CRBN of biotinylated proteins detected by LC-MS/MS analysis using IMR32 cells. Overlapping biotinylated proteins were compared between AirID-CRBN and TurboID-CRBN using a Venn diagram. f LC-MS/MS analysis of biotinylated peptides using AirID-CRBN or TurboID-CRBN in IMR32 cells. IMR32 cells stably expressing AGIA-AirID-CRBN or AGIA-TurboID-CRBN were treated with DMSO or 10 µM pomalidomide (Po) in the presence of 10 µM biotin and 5 µM MG132 for 2 h (biological replicates; n = 3). Significant changes in the volcano plots were calculated by Student’s two-sided t-test and the false discovery rate (FDR)-adjusted P-values calculated using Benjamini–Hochberg method are shown in the Supplementary Data 9, 10. a–d Biotinylated proteins were pulled down using streptavidin beads and analysed by immunoblotting. Source data are provided as a Source data file.

Fig. 6. ZMYM2-FGFR1 fusion protein is degraded by pomalidomide.

a Schematic diagram of ZMYM family proteins. b In vitro interaction assay between CRBN and ZMYM2, ZMYM3 or ZMYM4. The interaction between bls-CRBN and FLAG-GST-ZMYM2, -ZMYM3 or -ZMYM4 in the presence of DMSO or 20 µM thalidomide (Th), lenalidomide (Le), pomalidomide (Po) or 5-hydroxythalidomide (5HT) was analysed using the AlphaScreen-based biochemical assay. All relative AlphaScreen (AS) signals were expressed as a luminescence signal relative to that of DMSO. Error bars denote the standard deviation (independent experiments; n = 3). c In vitro interaction assay between CRBN and ZMYM2-WT or ZMYM2-C470A. The interaction between bls-CRBN and FLAG-GST-ZMYM2-WT or -C470A in the presence of DMSO or 20 µM pomalidomide (Po) was analysed using the AlphaScreen-based biochemical assay. All relative AlphaScreen (AS) signals were expressed as a luminescence signal relative to that of DMSO. Error bars denote the standard deviation (independent experiments; n = 3). d Immunoblot analysis of ZMYM2-AGIA protein levels in HEK293T cells expressing ZMYM2-WT- or ZMYM2-C470A-AGIA and FLAG-CRBN treated with DMSO or pomalidomide (Po) for 16 h. The experiments were repeated three times independently with similar results. e Schematic diagram of ZMYM2, FGFR1, ZMYM2-FGFR1 protein. f In vitro interaction assay between CRBN and ZMYM2, FGFR1 or ZMYM2-FGFR1. The interaction between bls-CRBN and FLAG-GST-ZMYM2, -FGFR1 or -ZMYM2-FGFR1 in the presence of DMSO or 20 µM pomalidomide (Po) was analysed using the AlphaScreen-based biochemical assay. All relative AlphaScreen (AS) signals were expressed as a luminescence signal relative to that of DMSO. Error bars denote the standard deviation (independent experiments; n = 3). g Immunoblot analysis of protein levels in HEK293T cells stably expressing empty, ZMYM2-WT- FGFR1- or ZMYM2-C470A-AGIA treated with DMSO or pomalidomide (Po) every 24 h for 48 h. The experiments were repeated three times independently with similar results. h Immunoblot analysis of protein levels in HEK293T cells stably expressing empty, ZMYM2-FGFR1-WT-AGIA (WT) or ZMYM2-C470A-AGIA (C470A) treated with DMSO or pomalidomide (Po) every 24 h for 48 h. The experiments were repeated three times independently with similar results. Source data are provided as a Source data file.

Fig. 7. AirID can be applied to other molecular glues and PROTACs.

a Chemical structure of indisulam. b Schematic diagram of indisulam-dependent biotinylation of RBM39 by AirID-DCAF15. c Indisulam-dependent biotinylation of RBM39 in HCT116 cells. HCT116 cells stably expressing AGIA-AirID-DCAF15-WT or -D475N were treated with DMSO or 5 µM indisulam and 50 µM biotin and 10 µM MG132 for 6 h. The experiments were repeated twice independently with similar results. d LC-MS/MS analysis of indisulam-dependent biotinylation in HCT116 cells. HCT116 cells stably expressing AGIA-AirID-DCAF15-WT or -D475N were treated with DMSO or 5 µM indisulam in the presence of 50 µM biotin and 10 µM MG132 for 6 h (biological replicates; n = 3). Significant changes in the volcano plots were calculated by Student’s two-sided t-test and the false discovery rate (FDR)-adjusted P-values calculated using Benjamini–Hochberg method are shown in Supplementary Data 11. e Chemical structures of dBET1 and ARV-825. f Pomalidomide (Po)- or PROTACs-dependent biotinylation assay of neo-substrates and BRD4 in MM1.S cells. MM1.S cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO, 1 µM Po, 10 µM dBET1 or 0.1 µM ARV-825 in the presence of 10 µM biotin and 5 µM MG132 for 8 h. The experiments were repeated three times independently with similar results. g Time course and dose-dependent analyses of PROTACs-induced biotinylation of neo-substrates and BRD4 in MM1.S cells. MM1.S cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO or ARV-825 (0.01, 0.1, or 1 µM) in the presence of 10 µM biotin and 5 µM MG132 for 2, 4 or 6 h. The experiments were repeated twice independently with similar results. h LC-MS/MS analysis of PROTACs-dependent biotinylation in MM1.S cells. MM1.S cells stably expressing AGIA-AirID-CRBN-WT were treated with DMSO, 1 µM pomalidomide (Po) or 0.1 µM ARV-825 in the presence of 5 µM biotin and 5 µM MG132 for 6 h (biological replicates; n = 3). Significant changes in the volcano plots were calculated by Student’s two-sided t-test and the false discovery rate (FDR)-adjusted P-values calculated using Benjamini–Hochberg method are shown in Supplementary Data 12. i Heat map of biotinylated peptides of known neo-substrates and ARV-825-targeted protein in MM1.S cells detected by LC-MS/MS analysis. c, f, g Biotinylated proteins were pulled down using streptavidin beads and analysed by immunoblotting. Source data are provided as a Source data file.