Monoubiquitination of ASXLs controls the deubiquitinase activity of the tumor suppressor BAP1

Abstract

The tumor suppressor and deubiquitinase (DUB) BAP1 and its Drosophila ortholog Calypso assemble DUB complexes with the transcription regulators Additional sex combs-like (ASXL1, ASXL2, ASXL3) and Asx respectively. ASXLs and Asx use their DEUBiquitinase ADaptor (DEUBAD) domain to stimulate BAP1/Calypso DUB activity. Here we report that monoubiquitination of the DEUBAD is a general feature of ASXLs and Asx. BAP1 promotes DEUBAD monoubiquitination resulting in an increased stability of ASXL2, which in turn stimulates BAP1 DUB activity. ASXL2 monoubiquitination is directly catalyzed by UBE2E family of Ubiquitin-conjugating enzymes and regulates mammalian cell proliferation. Remarkably, Calypso also regulates Asx monoubiquitination and transgenic flies expressing monoubiquitination-defective Asx mutant exhibit developmental defects. Finally, the protein levels of ASXL2, BAP1 and UBE2E enzymes are highly correlated in mesothelioma tumors suggesting the importance of this signaling axis for tumor suppression. We propose that monoubiquitination orchestrates a molecular symbiosis relationship between ASXLs and BAP1.

Fig. 1

ASXL2 is monoubiquitinated on DEUBAD in a BAP1-dependent manner. a Schema representation of ASXLs protein family, Asx and BAP1/Calypso proteins. b Cartoon representation of BAP1/Ub/DEUBAD (DEUBAD of ASXL2) homology structure, based on the UCH37/Ub/DEUBAD (DEUBAD of RPN13) crystal structure (PDB, 4UEL). c BAP1 enhances ubiquitination of ASXL2. HEK293T cells were transfected with Myc-ASXL2, Flag-BAP1 or HA-Ub vectors and subjected to immunoprecipitation and immunoblotting. n = 3 biological replicates. d ASXL2 is monoubiquitinated in BAP1-dependent manner. HEK293T cells were transfected as indicated and ASXL2 ubiquitination with GFP-Ub was analyzed by immunoblotting. n = 4 biological replicates. e Decrease of ASXL2 protein levels in U-2 OS cells stably expressing shRNA of BAP1. n = 5 biological replicates. f Depletion of ASXL2 using siRNA in U-2 OS cells. n = 5 biological replicates. g DEUBAD is required for ASXL2 monoubiquitination. HEK293T cells were co-transfected with Flag-BAP1 and the corresponding deletion mutants constructs of Myc-ASXL2 in presence or not of GFP-Ub and subjected to immunoblotting. n = 4 biological replicates. h The DEUBAD is sufficient for its monoubiquitination in BAP1-dependent manner. The indicated constructs were transfected in HEK293T cells and DEUBAD ubiquitination was analyzed. n = 4 biological replicates. i Mass spectrometry (MS) spectrum indicating Ub remnant on Lysine 370 of ASXL2. j Lysine 370 is the BAP1-dependent monoubiquitination site of ASXL2. HEK293T cells were transfected with the corresponding lysine mutants of DEUBAD (ASXL2) and used for immunoblotting. n = 3 biological replicates. k Conservation of the ASXLs ubiquitination site. Sequence alignment of ASXLs orthologs (ASXL2 K370 site highlighted in purple). l The DEUBAD of Asx is monoubiquitinated in Drosophila. S2 cells were transfected with Myc-V5-DEUBAD (Asx) and Myc-V5-DEUBAD (Asx) K325R expression vectors and subjected to immunoblotting. n = 3 biological replicates. m Drosophila DEUBAD K325 is monoubiquitinated in Calypso-dependent-manner. Myc-V5-DEUBAD (Asx) was co-transfected with control or Calypso dsRNA in S2 cells and its monoubiquitination levels were determined by immunoblotting. n = 3 biological replicates. Tubulin was used as a loading control for panels c–h and j. Histone H3 was used as a loading control for panels l, m

Fig. 2

Monoubiquitination of DEUBAD requires BAP1 intramolecular interactions. a Representation of the different BAP1 mutant forms used for experiments done in the panels b and d. b Disruption of the intra-molecular interactions between the catalytic (UCH) and non-catalytic (CC1 and CTD) domains of BAP1 as well as inhibition of its interaction with DEUBAD by the R666-H669 cancer associated mutation impair DEUBAD domain monoubiquitination. HEK293T cells were transfected with BAP1 and its mutant forms as indicated in presence of Myc-DEUBAD (ASXL2) and subjected to western blotting. n = 4 biological replicates. c Reconstitution of the intramolecular interactions of BAP1 promotes DEUBAD monoubiquitination. Co-transfection of Myc-DEUBAD (ASXL2) or Myc-DEUBAD (ASXL2) K370R with the corresponding GFP BAP1- or Myc-BAP1- fragments fusion constructs in HEK293T and cell extracts were used for western blotting. The star indicates a possible degradation product. The band upper UCH or UCH-CC1 represents a potential post-translational modification of this domain. n = 2 biological replicates. d Monoubiquitination of the DEUBAD domain following expression of BAP1^∆HBM, BAP1^∆UCH, or BAP1 catalytic dead (C91S) constructs in HEK293T cells. Increased amounts of the different Flag-BAP1 constructs were used in presence of HA-Ub and Myc-DEUBAD (ASXL2), then DEUBAD monoubiquitination was visualized by western blotting. n = 4 biological replicates. e Validation of Ub binding mutants of BAP1. Flag-BAP1 and its different mutant forms were transfected in HEK293T cells. Cell lysates were labeled with HA-Ub-VME DUB probe and analyzed by immunoblotting. n = 2 biological replicates. f BAP1 ubiquitin binding is not required for DEUBAD monoubiquitination. HEK293T cells were transfected with BAP1 Ub binding mutants in presence of Myc-DEUBAD (ASXL2) and harvested for immunoblotting. n = 2 biological replicates. Tubulin was used as a loading control for panels b–f

Fig. 3

Dual role of K370 ubiquitination on ASXL2 protein stability. a ASXL2 protein levels are accumulated following proteasome inhibition. U-2 OS or HEK293T cells were treated with 20 μM of MG132 for the indicated times and harvested for immunoblotting. n = 2 biological replicates. b HEK293T cells were treated with 20 μg/ml of cycloheximide (CHX) as indicated and ASXL2 levels were determined by immunoblotting. The arrows in panels a and b indicate the unmodified and the monoubiquitination forms of ASXL2. n = 2 biological replicates. c The non-modified form of ASXL2 is rapidly accumulated relative to its monoubiquitinated form following proteasome inhibition. U-2 OS cells were infected with lentiviral expressing vectors for ASXL2 or ASXL2 K370R then treated with 20 μM of MG132 for the indicated times and used for immunoblotting. n = 2 biological replicates. d Monoubiquitination of ASXL2 and mutation of K370 maintains its stability. U-2 OS cells stably expressing HA-ASXL2 or HA-ASXL2 K370R were treated with CHX as indicated and analyzed by immunoblotting as in c. n = 2 biological replicates. The arrows in panels c and d indicate the unmodified and the monoubiquitination forms of HA-ASXL2. e Monoubiquitination of DEUBAD and mutation of its K370 maintain its stability. U-2 OS cells stably expressing GFP-DEUBAD (ASXL2) or GFP-DEUBAD (ASXL2) K370R were treated with CHX as indicated. Cell lysates were used for immunoblotting. n = 3 biological replicates. f Unmodified DEUBAD is strongly accumulated relative to the monoubiquitinated form, following proteasome inhibition. U-2 OS cells stably expressing GFP-DEUBAD (ASXL2) or GFP-DEUBAD (ASXL2) K370R were treated with 20 μM of MG132 for the indicated times. Cell extracts were analyzed by immunoblotting. n = 3 biological replicates. g Further evidence that DEUBAD (ASXL2) K370 is also a site for Ub chain extension and degradation. HEK293T cells were transfected with Myc-DEUBAD (ASXL2) and HA-tagged Ub vectors and harvested for immunoblotting. n = 2 biological replicates. h ASXL2 stability is regulated by polyubiquitination inducing its degradation. HEK293T cells were transfected with Myc-ASXL2 and HA-tagged Ub vectors as indicated and subjected to immunoblotting. n = 2 biological replicates. Tubulin was used as a loading control for all panels

Fig. 4

K370 is the site of mono- and polyubiquitination. a DEUBAD protein levels are affected by DUB depletion. U-2 OS cells stably expressing Flag-HA-BAP1 and GFP-DEUBAD (ASXL2) were transfected with siRNA library for all human DUBs (see also Supplementary Fig. 3b). Densitometry analysis of protein bands in DUB siRNA versus non-target (NT) siRNA control were conducted. The arrows indicate the top DUBs hits that increased the corresponding protein levels. n = 1 biological sample. b Comparison of changes in band intensity between unmodified and monoubiquitinated forms of DEUBAD (ASXL2). c Validation that K370 is the site of mono- and poly-ubiquitination of DEUBAD (ASXL2). U-2 OS cells stably expressing Flag-HA-BAP1 and either GFP-DEUBAD (ASXL2) or GFP-DEUBAD (ASXL2) K370R were transfected with siRNA NT control or siRNAs targeting different DUBs and harvested for western blotting as indicated. n = 2 biological replicates. Tubulin was used as a loading control for panel c

Fig. 5

DEUBAD monoubiquitination is directly catalyzed by UBE2E family. a UBE2E1 and UBE2E3 directly catalyze monoubiquitination of DEUBAD (ASXL2). Bacterial purified His-BAP1/MBP-DEUBAD (ASXL2) complex was incubated with the indicated recombinant UBE2s conjugating enzymes for in vitro ubiquitination assays. The reactions were analyzed by western blot as indicated. The black dots indicate UBE2E1 and UBE2E3. n = 1 biological sample. b UBE2E3 in vitro monoubiquitinates K370 of DEUBAD (ASXL2) alone or DEUBAD (ASXL2) in complex with BAP1. Bacteria purified MBP-DEUBAD (ASXL2), MBP-DEUBAD (ASXL2) K370R, His-BAP1/MBP-DEUBAD (ASXL2) and His-BAP1/MBP-DEUBAD (ASXL2) K370R complexes were used for in vitro ubiquitination assays with bacteria-purified UBE2E3 and analyzed by immunoblotting. The arrows indicate the monoubiquitinated form of DEUBAD domain. n = 3 biological replicates. c Bacteria purified His-BAP1/MBP-DEUBAD (ASXL2) or His-BAP1 C91S/MBP-DEUBAD (ASXL2) complexes were used for in vitro ubiquitination assays. Reactions were stopped at the indicated times for immunoblotting. d In vitro ubiquitination assays were conducted using bacteria purified His-BAP1/MBP-DEUBAD (ASXL2) complex or MBP-DEUBAD (ASXL2) in complex with the corresponding BAP1 mutants. Reactions were done for the indicted times and analyzed by immunoblotting. n = 2 biological replicates for panels c and d

Fig. 6

Dual role of UBE2E family in regulating ASXL2 protein stability. a Depletion of UBE2E3 results in increased DEUBAD (ASXL2) protein levels. U-2 OS cells stably expressing GFP-DEUBAD (ASXL2) were transfected with a different siRNAs oligos for UBE2E3 and then treated with CHX as indicated. Right panel, densitometry analysis of DEUBAD protein bands was conducted and presented as indicated. n = 3 biological replicates. b Combined depletion of all three UBE2Es resulted in a stronger increase of DEUBAD (ASXL2) protein levels which become similar to those of GFP-DEUBAD K370R. U-2 OS cells stably expressing GFP-DEUBAD (ASXL2) were transfected with Non-target siRNA control or a combination of UBE2Es (UBE2E1, UBE2E2, UBE2E3) siRNAs, treated with CHX as indicated and used for immunoblotting. Right panel, densitometry analysis of the protein levels of DEUBAD bands were conducted and presented as in a. n = 3 biological replicates. c Opposite effects of UBE2Es on DEUBAD (ASXL2) and ASXL2 protein levels depending on BAP1 expression. Flag-UBE2Es expression constructs were transfected in HEK293T cells in the presence of Myc-DEUBAD (ASXL2) or Myc-ASXL2 expression vectors in presence or not of BAP1 and cells were harvested for immunoblotting. n = 3 biological replicates. d Depletion of UBE2E1, UBE2E3 and the combination of three UBE2Es resulted in decreased protein levels of endogenous ASXL2 and BAP1. U-2 OS cells were transfected with Non-target control (NT) or UBE2E siRNAs and then treated with CHX and used for immunoblotting. Bottom panel, densitometry analysis of ASXL2 and BAP1 levels was conducted and presented for each siRNA. n = 3 biological replicates. e UBE2Es-mediated DEUBAD monoubiquitination is conserved in Drosophila. S2 cells were transfected with dsRNA for Calypso or UBCD2 and Myc-V5-DEUBAD (Asx). DEUBAD monoubiquitination was evaluated by immunoblotting. Bottom panel, densitometry analysis of the protein levels of monoubiquitinated form versus non-modified form of DEUBAD (Asx). n = 3 biological replicates. The stars in panels a, b, and d indicate UBE2E2 band detected with UBE2E1 antibody. Tubulin was used as a loading control for panels a–d and histone H3 was used as a loading control for panel e. Error bars in panels a, b, d, e represents s.d. (mean ± SD)

Fig. 7

Monoubiquitination of DEUBAD K370 promotes BAP1 DUB activity. a Expression of DEUBAD (ASXL2) but not DEUBAD (ASXL2) K370R, results in reduced levels of H2Aub in BAP1-dependent manner. U-2 OS cells stably expressing empty vector or Flag-HA-BAP1 were transduced with lentiviral expressing vectors for either GFP-DEUBAD (ASXL2) or GFP-DEUBAD2 (ASXL2) K370R and endogenous level of H2Aub was analyzed by immunofluorescence. GFP DEUBAD (ASXL2) (green), H2Aub (red), DAPI (blue). n = 3 biological replicates. b DEUBAD induces reduction of H2Aub levels in BAP1 catalytic activity dependent manner. U-2 OS cells stably expressing Flag-HA-BAP1 C91S were infected with either GFP-DEUBAD (ASXL2) or GFP-DEUBAD (ASXL2) K370R and H2Aub levels were assessed (see also Supplementary Fig. 7). n = 2 biological replicates. c Expression of ASXL2, but not ASXL2 K370R, results in reduced levels of H2Aub. U-2 OS cells stably expressing Flag-HA-BAP1 were infected with lentiviral expressing vectors for either Myc-ASXL2 or Myc-ASXL2 K370R and H2Aub changes were evaluated by immunostaining. n = 2 biological replicates. The cells showing either decrease or no change of H2Aub levels were encircled in panels a–c. Scale bar: 10 μm for panels a–c. d Monoubiquitinated form of DEUBAD (ASXL2) strongly promotes BAP1 DUB activity comparatively to DEUBAD (ASXL2) K370R. Purified BAP1/DEUBAD complexes from HEK293T cells were used for in vitro DUB assays with Flag-H2A nucleosomes and analyzed by immunoblotting. n = 2 biological replicates. e Monoubiquitinated DEUBAD promotes deubiquitination of H2Aub in Drosophila. S2 cells were transfected with either Myc-V5-DEUBAD (Asx) or Myc-V5-DEUBAD (Asx) K325R and harvested for immunoblotting. Right panel, densitometry analysis of the levels of H2Aub is shown. Error bars represent s.d. (mean ± SD). n = 3 biological replicates. Histone H3 was used as a loading control for panels d and e

Fig. 8

Expression of Asx K325R results in haltere to wing transformation. a Effect of the expression of Asx WT or K325R mutant during Drosophila larval disc development. Discs were stained with anti-H2Aub (red), anti-Ubx (white) and DAPI (blue). GFP (green) indicates expression pattern of en-Gal4 and is delimited by dashed lines. Genotypes are indicated (see also Supplementary Fig. 8). Scale bar: 20 µm. n = 3 biological replicates. b Partial haltere to wing homeotic transformation associated with expression of Asx K325R. Phenotypes associated with the expression of Asx K325R in adult Drosophila. Left panel, images showing whole Drosophila animals and dissected halteres. Right panel, quantification of haltere to wing transformation phenotype. The arrows indicate the halteres. n = 3 biological replicates. Scale bar: 200 µm for images 1, 2 and 3. Scale bar: 100 µm for images 1′, 2′ and 3′. c Wing development defects following expression of Asx or Asx K325R. Scale bar: 200 µm. n = 3 biological replicates. WD wing imaginal disc, HD Haltere imaginal disc, MD mesothoracic leg imaginal disc

Fig. 9

Expression of ASXL2 K370R reduces mammalian cell proliferation. a Enforced expression of ASXL2 K370R decreases cellular proliferation. U-2 OS cells were transduced with different amounts of lentiviral suspensions produced using ASXL2 or ASXL2 K370R constructs. Cells were selected by puromycin and harvested for immunoblotting (top panel). Equal numbers of puromycin-selected cells were plated for colony formation assay (CFA) (bottom panel). n = 2 biological replicates. b The cells infected in a, were treated with nocodazole for FACS analysis at the indicated times. Note that (+2×) refers to transduction of cells with twice the amount of virus we normally use for Myc-ASXL2, and (−2×) refers to transduction of the cells with two times less the amount of viruses we normally use for ASXL2 370R. This adjustment was conducted to correct for the expression levels usually higher for ASXL2 K370R. n = 2 biological replicates. c, d Normal diploid fibroblast IMR90 cells were transduced with viral expression constructs for ASXL2 or ASXL2 K370R. Cells were selected by puromycin and equal numbers were plated for phase contrast pictures (c) or cell counts (d). Scale bar: 50 µm for panel c. n = 2 biological replicates. (Exp.1 and Exp.2). e siRNA depletion of ASXL2 decreases cellular proliferation. U-2 OS cells were transfected with NT siRNA control or siRNA for ASXL2. Equal numbers of puromycin-selected cells were plated for CFA (left panel). Cells were treated with nocodazole for FACS analysis (right panel). f siRNA depletion of UBE2E3 decreases cellular proliferation. U-2 OS cells were transfected with individual siRNA constructs as indicated. Cells were plated for viability measurement using MTT assay. n = 3 biological replicates. Error bars represent s.d. (mean ± SD). g, h Inactivation of UBE2E3 locus decreases cellular proliferation. Schematic representation for gRNAs targeting the UBE2E3 locus (g top panel). U-2 OS cells were transduced with different lentiviral CRISPR/Cas9 constructs, selected by puromycin and harvested for immunoblotting (g bottom panel). n = 3 biological replicates. Equal numbers of puromycin-selected cells were plated for CFA (h). n = 2 biological replicates. i The cells selected as in h were treated with nocodazole for FACS analysis at the indicated time . n = 2 biological replicates. j Positive correlation of BAP1, ASXL2, and UBE2E3 protein expression levels in human mesothelioma. Mesothelioma biopsies were immunostained for ASXL2, UBE2E3, or BAP1 (see Supplementary Fig. 11a and Supplementary Table 3). Pictures were taken at 100× magnification. Scale bar: 100 μm. Tubulin was used as a loading control for panels a and g

Fig. 10

Regulation of the BAP1/ASXL2 complexes by DEUBAD ubiquitination. ASXL2 is constitutively monoubiquitinated by UBE2Es on its DEUBAD domain. Interaction of monoubiquitinated ASXL2 with BAP1 leads to its stabilization and the subsequent activation of the DUB complex. Otherwise monoubiquitinated ASXL2 is targeted by other E3 Ub-ligases for ubiquitin chain extension and proteasomal degradation. UBE2Es also target ASXL2 already in complex with BAP1, possibly dynamically regulating its activity. DUBs might also regulate the stability of both free and complexed ASXL2. Mutation of ASXL2 lysine 370 or cancer mutations that abolish ASXL2-BAP1 interaction lead to defective monoubiquitination and subsequent loss of BAP1 DUB activity and tumor suppression