The Bap1 ^{tm1c(EUCOMM)Hmgu} mouse strain carries a conditional (floxed) allele of the Bap1 gene; it was generated by the Wellcome Trust Sanger Institute Mouse Genetics Project and Infrafrontier/EMMA (www.infrafrontier.eu) (51–53), and described in our recent work (49). The strain was bred to the transgenic line expressing Cre recombinase under the control of the B cell lineage specific promoter mb1-Cre (from Prof. Michael Reth, MPI für Immunbiologie und Epigenetik, Germany) (54), and to the B6.129P2(Cg)-Ighg1^tm1(cre)Cgn /J mouse strain (Jackson Labs: 010611, Cγ1-cre) expresses Cre recombinase from the endogenous immunoglobulin heavy constant gamma 1 locus (Ighg1) (55). As we previously described (49), the loxP sites flank exons 6-12 (ENSMUSE00000121807-00000121801) of the main ENSMUST00000022458.10 Bap1 transcript, and the Bap1 ^Δ allele is predicted to disrupt BAP1 protein coding sequence from amino acid 126 (out of 728), precluding full expression of the N-terminal UCH catalytic domain and all the downstream domains of the protein. All lines were on the C57BL/6 genetic background. Both male and female animals were used in experiments, sex-matched between test and control groups. Experiments were in accordance with the guidelines of the Canadian Council on Animal Care and protocol MCGL-2011-6029 approved by the McGill Animal Care Committee. Genotyping was performed in house using the DreamTaq DNA Polymerase (Thermo Fisher Scientific) and primers from IDT Technologies.

Mouse B cell line CH12F3, which is a widely used model for the study of B cell activation and class switching (56), was maintained at 0.5-2 x10⁶ cells/mL in RPMI-1640 (Thermo Fisher Scientific) with 10% Fetal Calf Serum (FCS, Thermo Fisher Scientific), 2mM L-Glutamine, 100μg/mL streptomycin, 100U/mL penicillin (Thermo Fisher Scientific), and 5% NCTC (Sigma-Aldrich).

Bap1 ^Δ/Δ CH12F3 cells were generated through CRISPR-Cas9 gene editing according to our previously described protocols (49). gRNAs were designed with http://crispr.mit.edu online tools (57): gRNA_Bap1_Exon4_Chr14:32,066,155: gcaaatggatcgaagagcgc, gRNA_Bap1_Exon5_Chr14: 32,066,654: ggcgtgagtggcacaagagt, and cloned into the pSpCas9(BB)-2A-GFP (PX458) plasmid (Addgene, #48138). Sequence-verified plasmids were introduced into CH12F3 cells through nucleofection using the Amaxa Cell Line Nucleofector Kit V (Lonza) according to the manufacturer’s protocol. Single GFP⁺ IgA^- cells were sorted on day-2, and expanded for further 14 days to generate single cell clones. Loss of BAP1 protein was validated with Western blotting and the following antibodies: anti-BAP1 (D7W7O, Cell Signaling Technology) and anti-β-Actin (D6A8, Cell Signaling Technology).

AlamarBlue fluorogenic assay (ThermoFisher Scientific) was performed according to the manufacturer’s protocols, as we previously described (49). Cells were seeded into 96-well plates at 10^5/mL, rested overnight, and the AlamarBlue reagent added at 10% (v/v). Fluorescence intensity was recorded at the 4-hour timepoint on the EnSpire Plate reader (Perkin Elmer), at the 560nm excitation and 590nm emission wavelengths. For the CFSE-dilution assay, 5 × 10⁶ cells were incubated in 2 μM CellTrace™ CFSE (ThermoFisher Scientific) in 1mL of PBS for 10 min at 37°C. The cells were washed in PBS with 5% FBS, followed by a second wash in PBS. The cells were re-suspended in complete media at 2 × 10⁵/mL and maintained in culture overnight at 37°C 5% CO₂. The cells were counterstained with Fixable Viability Dye eFluo780 (ThermoFisher Scientific), and the data were acquired on the BD Fortessa and analyzed with FlowJo (Tree Star, BD Biosciences) software. In some assays the CH12F3 cells were stimulated with TGF-β (1ng/ml, R&D Systems), IL-4 (10 ng/ml, Peprotech), and anti–CD40 (1μg/ml, BD Biosciences).

The protocols were as described in our recent work (58). Briefly, naïve primary mouse B cells were purified from splenocytes using EasySep™ Mouse B Cell Isolation Kit (19854, Stem Cell Technologies) and were cultured at 37°C with 5% CO₂ in iGB media: RPMI-1640 (Wisent), supplemented with 10% FBS (Wisent), 1% penicillin/streptomycin (Wisent), 0.1 mM 2-mercaptoethanol (BioShop), 10 mM HEPES, and 1 mM sodium pyruvate. Mouse primary B cells were labeled with 2.5μM CellTrace™ Violet (C34557, Invitrogen) in PBS for 20 minutes at 37°C before quenching as recommended by the supplier. For switching to IgA, B cells were stimulated with mouse recombinant IL-21 (20ng/mL, Peprotech), TGF-β1 (5ng/mL), retinoic acid (1μM, Sigma), F(ab’)2 goat anti-mouse IgM (5μg/mL, Jackson ImmunoResearch) and anti-CD40 mAb FGK45 (5μg/mL, prepared in-house from hybridoma), as previously described (59). To measure class switching to IgA, cells were treated with mouse FcR blocking reagent (130-092-575, Miltenyi Biotec) then stained with anti IgA-PE (1040-09, Southern Biotech). Dead cells were excluded using eBioscience™ Fixable Viability Dye eFluor™ 780 (65-0865-14, Invitrogen). Induced GC B cells (iGBs) were generated on 40LB feeder cells (a gift from Dr. Daisuke Kitamura, Tokyo University of Science, Tokyo, Japan) (60). One day prior to B cell plating, 40LB cells were treated with 10μg/mL mitomycin C (SKU M-1108, AG Scientific) at 0.5 × 10⁶ cells per mL in 10cm dishes with DMEM media supplemented with 10% FBS (Wisent) and 1% penicillin/streptomycin (Wisent). After treatment, the cells were washed 6 times with 10 mL of PBS and plated at 1.3×10⁵ cells per well in 0.5 mL (24-well plate). Subsequently, purified naive B cells were plated onto the 40LB feeders at 2x10⁵ cells in 1mL of iGB media, supplemented with 1 ng/mL IL-4 (214-14, Peprotech). On days 3, 4 and 5, the cells were either harvested for downstream analyses in PBS with 0.5% BSA and 2 mM EDTA or fed with 1 mL of fresh iGB media with 1 ng/mL IL-4. Class switching to IgG1 was analyzed by flow cytometry, treating with mouse FcR blocking reagent and staining the cells with anti-IgG1 PE (550083, A85-1, BD Pharmingen) and anti-IgM BV421 (562595, R6-60.2, BD Pharmingen). For the analyses of cell division, iGBs were stained with 2.5μM CellTrace™ Violet on the day of plating, according to the manufacturer’s protocol. Cell numbers for the iGBs growth curves were calculated using Countess 3 Automated Cell Counter.

For the phycoerythrin (PE) immunization the mice were injected subcutaneously with immunogen emulsion comprising 100 μl CFA (Thermo Fisher Scientific), 15 μg of R-PE (ProZyme, Cedarlane), and 100 μl PBS, vortexed for 45 min prior to the injection. For the immunization with sheep red blood cells (SRBCs), the mice were injected intravenously with 10⁹ SRBCs (Innovative Research IC10-0210, Cedarlane) in 300uls of PBS. Immunization protocols were previously described (24, 61).

ELISA analyses of total immunoglobulin levels in the serum of naive mice (or in the supernatants from B cell cultures) used rat anti-mouse capture antibodies IgM II/41, IgG1 A85-3, and IgG3 R2-38 (BD Biosciences, 1 μg/ml) followed by detection with goat anti-mouse IgG(H+L)-HRP (Southern Biotechnologies). For the analysis of total IgG2c levels capture antibody goat anti-mouse Ig (H+L) and detection antibody goat anti-mouse IgG2c-biotin (Southern Biotechnologies), followed by streptavidin-HRP (Thermo Fisher Scientific) were used. Purified mouse IgM, IgG1, IgG2c, and IgG3 isotype control antibodies (BioLegend) were used as standards for the calculation of immunoglobulin concentrations. In the ELISA assays for the detection of PE-specific antibodies, the plates were coated with R-PE (10 μg/ml, ProZyme, Cedarlane), and developed with biotin goat anti-mouse IgM, IgG1, IgG2c, or IgG3 (Southern Biotechnologies), followed by streptavidin-HRP (Thermo Fisher Scientific). All the ELISAs were developed with SuperAquaBlue substrate (Thermo Fisher Scientific), and data acquired at 405 nm on EnSpire 2300 plate reader (PerkinElmer). Throughout the ELISA procedure, PBS with 0.05% Tween-20 was used as the Wash Buffer, PBS with 1% bovine serum albumin (Wisent) as the Blocking Buffer, and PBS with 0.1% bovine serum albumin (Wisent) as the Assay Diluent.

The detection of SRBC-specific antibodies was performed by flow cytometry, as described previously (62). Briefly, mouse serum samples were pre-diluted 1:60 in PBS and incubated with 3x 10⁵ SRBC for 30 min on ice. The SRBCs were washed and stained with goat anti-mouse IgG AlexaFluor488 (Poly4053, BioLegend, 1:200), or with goat anti-mouse IgM-biotin (1020-08, Southern Biotechnologies, 1:750), or IgG1-biotin (1070-08, Southern Biotechnologies, 1:750), followed by streptavidin – PerCP-Cy5.5 (BioLegend). Following thorough washing, the samples were analyzed on BD Fortessa and data processed with FlowJo software (Tree Star, BD Biosciences). SRBCs pre-incubated with serum of naive mice and stained as described above were analyzed as negative controls.

Cell suspensions of mouse tissues were prepared in RPMI-1640 (Thermo Fisher Scientific) with 2% (v/v) FCS, 100μg/ml streptomycin and 100U/ml penicillin (Thermo Fisher Scientific). The cells were stained for surface-markers in PBS with 2% FCS for 20 min on ice, with the following antibodies, as summarized in Supplementary Table S1 : Alexa Fluor 488 anti-GL7 (GL7, Biolegend); APC anti-CD21/CD35 (7E9, Biolegend), anti-CD86 (GL-1, Biolegend), anti-CD267/TACI (ebio8F10-3, eBioscience), and anti-IgD (11-26c.2a, BioLegend); APC-eFluor780 anti-CD45R/B220 (RA3-6B2, ThermoFisher); APC-Cy7 anti-CD5 (53-7.3, Biolegend); Brilliant UltraViolet 395 anti-CD43 (S7, BD Biosciences); Brilliant Violet 421 anti-CD95/Fas (Jo2, BD Biosciences) and anti-CD138 (281-2, Biolegend); Brilliant Violet 650 anti-CD45R/B220 (RA3-6B2, BioLegend); Biotin anti-IgG1 (polyclonal, 1070-08, Southern Biotech); eFluor450 anti-CD45R/B220 (RA3-6B2, Thermo Fisher); FITC anti-CD23 (B3B4, Invitrogen) and anti-Ki67 (SolA15, eBioscience); Pacific Blue anti-IgD (11-26c.2a, BioLegend); PE anti-Blimp-1 (5E7, Biolegend), anti-CD184/CXCR4 (L276F12, Biolegend) and anti-IgM (II/41, Thermo Fisher); PE-Cy7 anti-CD19 (6D5, BioLegend), anti-CD21/CD35 (eBio8D9, Invitrogen), and anti-CD38 (90, Invitrogen); PerCP-Cy5.5 anti-CD4 (RM4-4, Biolegend), anti-CD8a (53-6.7, Biolegend), anti-CD11b (M1/70, eBioscience), anti-CD19 (1D3, Tonbo Biosciences), anti-CD45R/B220 (RA3-6B2, BioLegend), anti-IgD (11-26c.2a Biolegend), anti-CD93 (AA4.1, Biolegend), anti-NK1.1 (PK136, Biolegend), and anti-TER119 (Ly-76, Biolegend). Brilliant Violet 785 Streptavidin (Biolegend) was used to detect biotin-conjugated antibodies. Viability Dye eFluor® 506 (eBioscience) was used to discriminate dead cells. Compensation was performed with BD CompBeads (BD Biosciences). The data were acquired on BD Fortessa and analyzed with FlowJo software (Tree Star, BD Biosciences).

Total B cells were isolated from mouse spleens via magnetic enrichment using with the EasySep Mouse CD19 Positive Selection Kit II (Stem Cell Technologies). For the FACS-sorting of B cell subsets, mouse spleens were mechanically dissociated in PBS with 0.1% BSA and 2mM EDTA, passed through 40 μm cell-strainers, and subjected to red blood cell lysis in ACK buffer (0.15M NH4Cl, 10mM KHCO3, 0.1mM EDTA). Cells were stained with the following antibodies, as summarized in Supplementary Table S2 : Alexa Fluor 488 anti-GL7 (GL7, Biolegend), Brilliant Violet 421 anti-CD95/Fas (Jo2, BD Biosciences), and Brilliant Violet 650 anti-CD45R/B220 (RA3-6B2, BioLegend). Viability Dye 7-AAD (Biolegend) was used to discriminate dead cells. Sorting was performed on the FACSAria (BD Biosciences).

ChIP was performed as described previously (49, 63). Cells were fixed with 1% formaldehyde in cell culture media for 10 min at room temperature, followed by addition of 0.125 M glycine to stop the fixation. Nuclei were extracted with 5 min lysis in 0.25% Triton buffer, followed by 30 minutes in 200mM NaCl buffer. Nuclei were resuspended in sonication buffer and sonicated for 12 cycles of 30 sec with a digital sonifier (Branson Ultrasonics) at 80%, with 30 sec rest in cooled circulating water. Beads immunocomplexes were prepared overnight by conjugating 40μL of Dynabeads Protein G (Thermo Fisher Scientific) with antibodies anti-BAP1 (Cell Signaling Technology, D1W9B, 52.8 μg) or anti-H2AK119ub (Cell Signaling Technology, D27C4, 5 μg). Immunoprecipitation was performed with an overnight incubation of the antibody-bead matrices with sheared chromatin from the equivalent of 5x10⁶ cells. Six washes were performed with low-stringency buffers, and samples de-crosslinked by overnight incubation in 1% SDS buffer at 65°C. Following treatments with RNaseA and Proteinase K, ChIP DNA was purified using Qiaquick PCR Cleanup kit (Qiagen).

ChIP-seq libraries were prepared using the Illumina TruSeq kit and sequenced on the Illumina NovaSeq 6000 in a paired-end 100bp configuration, with input DNA from the same cells sequenced as the negative controls. The reads were mapped to the UCSC mouse mm9 reference genome with Bowtie 1.0.0 (64), and BAP1 binding sites identified using the MACS1.4 peak detection algorithm (65), comparing for read enrichment against input DNA from the same cells. Normalized sequence read density profiles (bigwig) were generated with Homer (66) and visualized with IGV (67). Gene ontology (GO) enrichment analysis on the genes associated with the BAP1 ChIP-Seq binding clusters was performed on GREAT 4.0.4 (68) with Basal plus extension, searching for genes within 2kb upstream, 2kb downstream, or 200kb in distal.

RNA-seq protocols were as previously described (49, 69). Briefly, RNA was isolated using the Mag-MAX total RNA kit (Ambion) and quality assessed using Bioanalyzer RNA Pico chips (Agilent). rRNA depletion and library preparation were performed using the SMARTer Stranded RNA-Seq kit (Takara Clontech). The libraries were sequenced on an Illumina Novaseq 6000 in a paired-end 100 bp configuration aiming for 50x10⁶ reads per sample. The quality of the sequencing reads was confirmed using the FastQC tool (Babraham Bioinformatics), and low-quality bases were trimmed from the read extremities using Trimmomatic v.0.33 (70). The reads were then mapped to the mouse UCSC mm9 reference genome assembly using Hisat2 v2.2.1 (71–73). Gene expression was quantified by counting the number of uniquely mapped reads with featureCounts using default parameters (74). We retained genes that had an expression level of at least 5 counts per 10⁶ reads (CPM) in at least 4 of the samples (75). TMM normalization and differential gene expression analyses were conducted using the edgeR Bioconductor package (76). Dimension reduction analysis was performed using the Principal Component and Partial Least Squares regression method (77). Pairwise comparisons were performed between genotypes or between treatments, and genes with changes in expression ≥ |2.0| fold and Benjamini-Hochberg adjusted p values ≤ 0.05 were considered significant. Gene ontology (GO) enrichment analyses on differentially expressed gene clusters were performed with DAVID Bioinformatics Resources 6.8 (78), and Gene Set Enrichment Analysis (GSEA) was performed with GSEA tool v4.3.2 using MSigDB v2022.1 with default configuration and permutation within gene sets (79). For the RNA-Seq data consolidation with the ChIP-Seq data, full gene annotations were obtained from UCSC mouse mm9 reference genome. An in-house Python script was developed to load the genomic locations of ChIP-Seq binding sites and RNA-Seq dysregulated genes, and search for gene TSS within a specific distance to each ChIP-Seq binding site, as previously described (80).

Statistical analyses used Prism 9.5.1 (GraphPad Inc.), with Student’s t-test for two datasets and ANOVA for multiple comparisons, with further information is provided in Figure Legends.

Due to the high expression of Bap1 throughout the B cell lineage ( Supplementary Figure S1 ) (50, 81) and its important role as a DUB for H2AK119ub in other cell types (28), we hypothesized that BAP1 may act as an important regulator of the transcriptional programs of B cell mediated immune response. In order to test this, we adopted the previously established Bap1 ^fl/fl mb1-Cre mouse model, with the selective loss of BAP1 throughout the B cell lineage (49). We observed a strong reduction in total antibody titers in the serum of naïve Bap1 ^fl/fl mb1-Cre mice, relative to control mice of Bap1 ^fl/+ and Bap1 ^fl/+ mb1-Cre genotypes, including both IgM antibodies and class-switched IgG1, IgG2c, and IgG3 antibodies ( Figures 1A, B ). We proceeded to challenge the mice with subcutaneous injections of phycoerythrin (PE) in CFA adjuvant and observed a strong reduction in the antigen specific antibody responses in Bap1 ^fl/fl mb1-Cre relative to control mice, following both primary and boost immunizations, and affecting both IgM and class-switched IgG1, IgG2c and IgG3 isotypes ( Figures 1C–E ). Overall, this indicates the importance of BAP1 expression in the B cell lineage for the induction of B cell mediated immune response.

Bap1 ^fl/fl mb1-Cre mice lack BAP1 expression throughout the B cell lineage, resulting in impaired B cell development and reduced B cell numbers in lymphoid organs (49), which could in part account for the impaired antibody production in these mice. Our objective therefore was to test the direct role of BAP1 in the regulation of B cell mediated immune response, independently of its functions in B cell development. For this purpose, we generated the Bap1 ^fl/fl Cγ1-cre mice, with Cre expression driven from the IgG1 heavy chain locus and Cre activity restricted to the activated, germinal center (GC), and downstream B cell populations (55). In contrast to the Bap1 ^fl/fl mb1-Cre mice (49), naïve Bap1 ^fl/fl Cγ1-cre mice had normal numbers of splenic pre-GC B cells, including the transitional, follicular, and marginal zone subsets ( Supplementary Figure S2 ). At the same time, the effective Cre-mediated Bap1 ^fl to Bap1 ^Δ allele conversion was demonstrated by PCR-genotyping of the genomic DNA from ex vivo stimulated Bap1 ^fl/fl Cγ1-cre and control Bap1 ^+/+ Cγ1-cre B cells ( Supplementary Figures S3A, B ). Furthermore, RNA-seq analyses of splenic Bap1 ^fl/fl Cγ1-cre GC B cells demonstrated a strong reduction in the transcript reads mapping to the Bap1 floxed exons 6-12, as compared to the GC B cells from mice of control Bap1 ^+/+ Cγ1-cre genotype ( Supplementary Figure S3C, D ).

Cohorts of naïve Bap1 ^fl/fl Cγ1-cre mice were analyzed for total serum antibody levels and demonstrated a significant reduction in IgG1 but not IgM titers relative to the Bap1 ^+/+ Cγ1-cre control group ( Figure 2A ). Further cohorts of Bap1 ^fl/fl Cγ1-cre and control Bap1 ^+/+ Cγ1-cre mice were immunized, either subcutaneously with PE/CFA or intravenously with sheep red blood cells (SRBCs) and analyzed for antigen specific antibody titers over a time-course post-prime and post-boost immunization. A strong reduction in the antigen-specific antibody levels in Bap1 ^fl/fl Cγ1-cre mice relative to the control Bap1 ^+/+ Cγ1-cre group was observed across both immunization models and most time-points ( Figures 2B–E ). Overall, these findings demonstrate the direct and cell intrinsic role of BAP1 in the induction of B cell mediated immune response, independent of its previously reported functions in B cell development (49).

To understand the impact of BAP1-loss on the cellular dynamics of humoral immune response, mice of Bap1 ^fl/fl Cγ1-Cre and control Bap1 ^+/+ Cγ1-Cre genotypes, both naïve and challenged with SRBC-immunization (as shown in Figure 2C ), were analyzed for the numbers of germinal center (GC) B cells, memory B cells, and plasma cells using flow cytometry ( Figure 3 ). While there was no significant change in the absolute numbers of GC B cells in Bap1 ^fl/fl Cγ1-Cre relative to control mice at day 11 following the SRBC-immunization ( Figure 3A ), we observed a significant reduction in the ratio of dark zone (DZ) to light zone (LZ) GC B cells ( Figure 3B ), primarily due to an expansion in the number of LZ B cells (data not shown). Importantly, Bap1 ^fl/fl Cγ1-Cre mice showed a reduction in the numbers of class switched IgG1⁺ GC B cells ( Figure 3C ), with a significant depletion of IgG1⁺ cells within both the DZ and LZ subpopulations (data not shown), as well as a reduction in IgG1⁺ memory B cells ( Figure 3D ). Absolute numbers of plasma cells were also significantly reduced in both the spleen and bone marrow of Bap1 ^fl/fl Cγ1-Cre mice ( Figure 3F ), with a particularly strong depletion of the IgG1⁺ class-switched plasma cells ( Figure 3G ). Further analysis of the plasma cell population demonstrated a more significant depletion of plasmablasts and early plasma cells as compared to late plasma cells in the Bap1 ^fl/fl Cγ1-Cre mice ( Supplementary Figures S4A, B ), and a reduction in the proportion of plasma cells positive for the Ki-67 marker of cell proliferation at least in the bone marrow ( Supplementary Figure S4C ). Overall, the significant depletion of plasma cells likely contributes to the failure to produce and sustain normal antibody titers in the Bap1 ^fl/fl Cγ1-Cre mice.

The mice were further analyzed for GC B cells, memory B cells, and plasma cells at day 7 after the boost SRBC immunization ( Supplementary Figure S5 ). Surprisingly, an increase in GC B cells was observed in Bap1 ^fl/fl Cγ1-Cre relative to control mice after the booster immunization ( Supplementary Figure S5A ), and this may reflect the modulatory effects of pre-existing high antibody titers on B cell recruitment into the GC in the control group (82). Furthermore, numbers of IgG1⁺ class-switched GC and memory B cells in the Bap1 ^fl/fl Cγ1-Cre mice were restored to normal after the booster immunization ( Supplementary Figures S5C, D ), consistent with a considerable rise in their antibody titers at this time point ( Figures 2D, E ). However, the reduced ratio of DZ to LZ GC B cells and the significant depletion of class switched IgG1⁺ plasma cells persisted even after the booster immunization in Bap1 ^fl/fl Cγ1-Cre mice ( Supplementary Figures S5B, G ). The persisting plasma cell phenotype likely contributes to the ongoing reduction in the antibody titers in Bap1 ^fl/fl Cγ1-Cre mice.

While the more significant impact of BAP1-loss on the numbers of IgG1⁺ rather than total GC B cells and plasma cells in the Bap1 ^fl/fl Cγ1-Cre model may reflect the higher expression of Cre in the IgG1⁺ cells (55), it may also suggest a role for BAP1 in immunoglobulin class switching. We therefore proceeded to directly test the effects of Bap1-loss on antibody class switching in vitro. Induced GC B cells (iGBs) of Bap1 ^fl/fl Cγ1-Cre and control Bap1 ^+/+ Cγ1-Cre genotypes were generated on the 40LB feeder cells that provide CD40L and BAFF, as described previously (58, 60). iGBs were analyzed at day 4 for class switching to IgG1 and demonstrated no significant differences between the Bap1 ^fl/fl Cγ1-Cre and control Bap1 ^+/+ Cγ1-Cre cultures ( Figures 4A, B ). We further analyzed B cell proliferation in the iGBs cultures, using either the CellTrace Violet (CTV) dilution method with flow cytometry ( Figures 4C, D , day 4) or manual cell counting ( Figure 4E , days 0-6). This demonstrated a significant impairment in the proliferation for Bap1 ^fl/fl Cγ1-Cre relative to control Bap1 ^+/+ Cγ1-Cre B cells under these culture conditions ( Figures 4C–E ). Further analyses of the antibody class switching rates per cell division in this model confirmed that the loss of Bap1 does not directly interfere with class switching, and any reduction in class switched Bap1 ^fl/fl Cγ1-Cre B cells can be attributed to impaired B cell proliferation ( Figures 4C, F ).

Primary B cells from Bap1 ^fl/fl Cγ1-Cre and control Bap1 ^+/+ Cγ1-Cre mice were also stimulated in culture to class-switch to IgA with IL-21, anti-CD40, anti-IgM, TGF-β, and retinoic acid, as reported previously (59). Again, Bap1 ^fl/fl Cγ1-Cre B cells demonstrated normal levels of class switching over 4 days in culture ( Supplementary Figures S6A, B ), though Cre-mediated Bap1 ^fl to Bap1 ^Δ allele conversion was observed as early as days 3, albeit without full loss of the Bap1 ^fl allele ( Supplementary Figures S6C, D ). It is notable that no loss of cell proliferation was observed in these cultures, and this may reflect biological differences in the responses of Bap1-deficient B cells to different stimulation conditions or the lower efficiency of Cγ1-Cre in the cells undergoing class switching to IgA relative to IgG1, as shown by the persistent detection of the Bap1 ^fl allele in these cells ( Supplementary Figure S6D ).

We further evaluated class switching using splenic B cells from Bap1 ^fl/fl mb1-Cre and control Bap1 ^+/+ mb1-Cre mice, as in this model the Cre-mediated Bap1-allele deletion takes place from the early stages in B cell lineage development, and the full loss of BAP1 protein in Bap1 ^fl/fl mb1-Cre splenic B cells has been validated through Western blotting in our previously published studies (49). Thus, to further assess the role of BAP1 in antibody class switching, Bap1 ^fl/fl mb1-Cre and control Bap1 ^+/+ mb1-Cre B cells were stimulated in culture with IL-21, anti-CD40, anti-IgM, TGF-β, and retinoic acid to class-switch to IgA and analyzed at day 3 ( Figures 4G–K ). While Bap1 ^fl/fl mb1-Cre B cells showed a trend toward impaired cell proliferation relative to the control B cells ( Figures 4G, I, J ), there were no differences in class switching per cell division between the Bap1-genotypes ( Figure 4K ), consistent with our findings with the Bap1 ^fl/fl Cγ1-Cre B cells ( Figure 4F , Supplementary Figure S6B ). Overall, we conclude that BAP1 is not directly required for the normal progression of antibody class switching.

Given the impaired proliferation of Bap1-deficient B cells under some of the stimulation conditions presented above, we further analyzed the effects of BAP1 on B cell viability. No significant differences in cell viability were observed between Bap1 ^fl/fl Cγ1-Cre and control Bap1 ^+/+ Cγ1-Cre germinal center B cells, memory B cells, or plasma cells, freshly isolated from naive mice or from mice at day 11 post-SRBC immunization ( Supplementary Figures S7A–D ). Further analysis of the plasma cell populations as plasmablasts, early plasma cells, and mature plasma cells also demonstrated no differences in viability based on the Bap1-genotype of the cells (data not shown). However, such analyses may not detect subtle changes in cell viability due to the rapid clearance of dead cells in healthy tissues, and we therefore further analyzed the viability of primary B cells of Bap1 ^fl/fl Cγ1-Cre and control Bap1 ^+/+ Cγ1-Cre genotypes in culture, stimulated over 5 days with either anti-CD40 and IL-4, or with LPS and IL-4. This demonstrated a mild but statistically significant reduction in the viability of Bap1 ^fl/fl Cγ1-Cre relative to control Bap1 ^+/+ Cγ1-Cre B cells under both stimulation conditions ( Supplementary Figure S7E ). Overall, we conclude that the loss of BAP1 can impair the viability of activated B cells.

To further explore the roles and mechanisms of BAP1 activity in B cells, we targeted Bap1 gene using CRISPR/Cas9 in the CH12F3 B cell lymphoma line (56), which is a commonly used model for the analyses of B cell activation, class switching, and associated cellular processes. gRNAs were designed to target Bap1 exons 4 and 5 that encode the N-terminal catalytic domain of the protein ( Figure S8A ), and single-cell clones screened by PCR for deletions within the targeted regions of the Bap1 locus. Further analyses of five Bap1 ^Δ/Δ single-cell clones by Western blotting confirmed the full loss of BAP1 protein expression ( Supplementary Figure S8B ).

Based on the impaired proliferation of Bap1 ^fl/fl Cγ1-Cre primary B cells in the iGB cultures ( Figure 4 ) and on our previous report of impaired proliferation in Bap1 ^fl/fl mb1-Cre pre-B cells (49), we analyzed the Bap1 ^Δ/Δ CH12F3 cell clones for defects in proliferation, using the AlamarBlue resazurin/resorufin-based assay in bulk cell cultures, and the flow cytometry based CFSE dilution method. The assays demonstrated a significant reduction in cell proliferation for the Bap1 ^Δ/Δ CH12F3 cell clones relative to control wild type CH12F3 cells ( Supplementary Figures S8C–E ), as well as the reduced viability of Bap1 ^Δ/Δ CH12F3 cell at least under unstimulated culture conditions ( Supplementary Figure S8F ). This demonstrates the conservation of Bap1-deficiency phenotypes between CH12F3 cells and primary B cells and establishes the Bap1 ^Δ/Δ CH12F3 cells as a model for further analyses of BAP1 functions in B cell biology.

To identify the genomic regions directly regulated by BAP1 in mature B cells, we conducted ChIP-seq to map BAP1 genome-wide DNA-binding sites using an anti-BAP1 antibody and wild type CH12F3 cells, with and without stimulation with TGF-β, IL-4, and anti-CD40 over 72 hours. While there are limitations to using CH12F3 cells as a model of primary GC B cells, BAP1 is a challenging target, with most previously published ChIP-seq datasets using cells that express an epitope-tagged BAP1 protein (47, 83, 84). We have therefore also repeated the ChIP-seq analysis using CH12F3 cells stably expressing 3xFLAG-tagged BAP1 with an antibody against the FLAG epitope, as previously described (47, 49). We identified a total of 1,499 BAP1 binding sites (or peaks) across the genome ( Figure 5A , Supplementary Figure S9A , Supplementary Table S3A ), with the high concordance between the dataset from unstimulated and stimulated CH12F3 cells, indicating that BAP1 recruitment to chromatin is constitutive rather than induced with B cell stimulation. Among the BAP1 binding sites 88% were classified as gene-proximal based on their distance to the nearest gene transcription start site (≤1kb to TSS, Supplementary Table S3A ). To gain insights into the biological functions of the BAP1 transcriptional target genes, ontology analyses were performed on the genes in the vicinity of the BAP1 binding sites using the Genomic Regions Enrichment of Annotations Tool (GREAT) (68). This demonstrated a highly significant association of the gene-proximal BAP1 binding sites with the genes involved in proteosome-dependent protein degradation, DNA damage response, and protein trafficking, and the association of the gene-distal BAP1 binding sites with the genes involved in the regulation of transcription and RNA processing ( Figure 5B , Supplementary Tables S3B, C ). This suggests the role of BAP1 in the regulation of many transcriptional programs essential for normal cell physiology, as well as for the induction of B cell activation and humoral immune response.

To compare the role of BAP1 as a transcriptional regulator in mature B cells to its role in other cell types, we consolidated our newly acquired BAP1 ChIP-Seq with the previously published BAP1 ChIP-Seq datasets from Ba/F3 pro-B cells (49), ES cells (85), and macrophages (47). This indicated a considerable overlap in BAP1 genomic binding and suggested shared BAP1 functions in the regulation of housekeeping transcriptional programs across many cell types ( Supplementary Figures S9A , Supplementary Tables S3D, E ). To gain further insights into the cross-talk between BAP1 and other transcriptional regulators, we also consolidated our BAP1 ChIP-Seq data with the public ChIP-Seq datasets for BAP1-associated transcriptional regulators, including HCF1 and OGT in macrophages (47), and ASXL1 in hematopoietic stem and progenitor cells (HSPC) (86). A significant overlap in the genomic localization was observed between BAP1 and BAP1-interacting proteins ASXL1, HCF1, and to a lesser extent OGT ( Supplementary Figure S9B , Supplementary Tables S3D, E ), indicating that BAP1 may act in cooperation with these transcriptional regulators in B cells, as in other cell types. Furthermore, we consolidated our BAP1 ChIP-seq data with the public ChIP-seq datasets from primary murine B cells for the other major regulators of the histone H2AK119ub epigenetic mark, including polycomb proteins RING1B, CBX7, YY1, EZH2 and deubiquitinase USP16 (18, 27). We observed an overlap in the genomic binding of BAP1 with components of the polycomb repressive complex 1 (PRC1), including its E3 ubiquitin ligase catalytic subunit RING1B ( Supplementary Figure S9B , Supplementary Table S3D ). This suggests that BAP1 and PRC1 may co-regulate histone H2AK119ub levels and turnover at shared genomic locations. Interestingly, we also observed an overlap in the genomic binding of BAP1 and another well-known H2AK119ub-deubiquitinase USP16 ( Supplementary Figure S9B , Supplementary Table S3D ), suggesting their cooperative or complementary functions.

To assess the role of BAP1 as a DUB for histone H2AK119ub in activated B cells, consistent with its reported function in other cell types (28), further ChIP-Seq analyses were performed to quantify the histone mark H2AK119ub. The study compared control wild type (WT) and BAP1-deficient Bap1 ^Δ/Δ CH12F3 B cell lines, analyzing three independent clones of CH12F3 cells (KO3-4-5) and repeating the analyses at steady-state and following cell activation (TGFβ, IL-4, anti-CD40, 72 hours). ChIP-Seq data demonstrated that the loss of BAP1 resulted in a highly significant increase in histone H2AK119ub levels at the BAP1 binding sites and surrounding genomic regions ( ± 5kb) in all the three Bap1 ^Δ/Δ clones and under both untreated and stimulated conditions ( Figures 5C ). This establishes the non-redundant role of BAP1 as a DUB for histone H2AK119ub in B lymphocytes.

Next, we set out to interrogate the biological functions of the genes regulated via the BAP1 DUB-activity for H2AK119ub. We therefore re-analyzed our ChIP-seq datasets to identify the genes with a >2-fold increase in H2AK119ub within ±1.5kb to their TSS in the Bap1 ^Δ/Δ relative to control CH12F3 B cells ( Supplementary Table S4A ). GO-term analysis on this gene-set indicated that the genes undergoing an increase in H2AK119ub with the loss of BAP1 encode regulators of protein transport, Ub-dependent protein degradation, DNA damage response and repair, RNA processing, as well as autophagosome assembly and cell cycle progression ( Figure 5D , Supplementary Table S4B ). This demonstrates a strong concordance in the biological functions of this gene-set with the genes marked by BAP1-binding ( Figures 5B, D ). Overall, we conclude that BAP1 acts as a DUB for H2AK119ub in B cells to regulate many transcriptional programs essential for normal cell physiology, but also for B cell activation and the induction of humoral immune response.

For further insights into BAP1 functions as a DUB for H2AK119ub in the transcriptional networks of B cell mediated immune response, we performed supplementary analyses to consolidate our ChIP-seq datasets with the published RNA-seq data characterizing the transcriptional response of wild type CH12F3 cells to stimulation (87), and the transcriptional change with the transition from follicular to GC in primary B cells (81). In both cases, we identified a subset of differentially expressed genes that in our datasets have proximal BAP1 binding and sustain an increase in H2AK119ub with the loss of BAP1 ( Supplementary Tables S4D-F ). Such datasets may represent a starting point for future studies to validate and characterize genes under the direct transcriptional control of BAP1 in activated B cells.

To further analyze the role of BAP1 in the transcriptional regulation of B cell activation, RNA-Seq gene expression analysis was conducted on GC B cells isolated from the spleen of Bap1 ^fl/fl Cγ1-cre and control Bap1 ^+/+ Cγ1-cre mice at day 11 post-SRBC immunization, with the cells gated for FACS-sorting as shown in Supplementary Figure S10 . As already discussed, a strong reduction in the expression of the Bap1 ^fl exons was demonstrated in the RNA-seq datasets from Bap1 ^fl/fl Cγ1-cre relative to control GC B cells ( Supplementary Figures S3C, D ), confirming an effective Cre-mediated Bap1 gene inactivation in the cells captured for the RNA-seq. Importantly, dimension reduction analysis of the gene expression profiles revealed a clear segregation of the Bap1 ^fl/fl Cγ1-cre and control Bap1 ^+/+ Cγ1-cre GC datasets ( Figure 6A , Supplementary Table S5 , PC1 – 31% and PC2 –14% variance), which indicates a significant transcriptional change in GC B cells with the loss of BAP1.

To explore the biological functions of the genes dysregulated in Bap1-deficient GC B cells, gene set enrichment analysis (GSEA) was performed (79). This demonstrated a significant downregulation of the transcriptional signatures involved in B cell engagement with T-helper cells, such as ‘positive regulation of T cell proliferation’ and ‘positive regulation of IL2 production’ in the Bap1 ^fl/fl Cγ1-cre GC B cells, as well as an upregulation of the transcriptional signatures linked to mitochondrial function such as ‘inner mitochondrial membrane organization’, ‘ATP biosynthetic process’, and ‘oxidative phosphorylation’ ( Figure 6B , Supplementary Table S5C ). We also conducted a differential gene expression analysis between the Bap1 ^fl/fl Cγ1-cre and control Bap1 ^+/+ Cγ1-cre GC B cells, at the fold change (FC) ≥ |2.0| and false discovery rate (FDR) ≤ 0.05, which identified 94 upregulated and 76 downregulated genes in the Bap1 ^fl/fl Cγ1-cre GC B cells ( Supplementary Table S5A ). Applying gene ontology (GO) analysis, we note the downregulation of the transcriptional signature of ‘intracellular signal transduction’ in the Bap1 ^fl/fl Cγ1-cre GC B cells ( Figure 6C , Supplementary Table S5E ), with the downregulated genes including important positive and negative regulators of B cell receptor signaling, such as Nfam1 (88), Camk2d (89), Lrrk2 (90), Dgka (91), and Cblb (92, 93) ( Figure 6D ). As an example, Cblb encodes an E3 ubiquitin ligase that targets CD79a, CD79b, and IRF4 to regulate the initiation and progression of the GC-reaction (92, 93). Overall, the cell intrinsic loss of BAP1 in GC B cells resulted in a dysregulation of the transcriptional programs essential for normal induction of humoral immune response.

We consolidated the RNA-seq and ChIP-seq datasets to identify putative genes under the direct transcriptional control of BAP1 in B cells, as the genes dysregulated in Bap1 ^fl/fl Cγ1-cre GC B cells and with BAP1-binding at their promoters in CH12F3 and/or BaF3 B cell lines. While our datasets showed no evidence for BAP1 regulation of the genes encoding the major mediators of B cell lineage choice, GC reaction, or plasma cell differentiation (Bach2, Batf, Bcl6, Blimp1/Prdm1, Irf4, Irf8, Myc, Pax5, Xbp1, Supplementary Tables S3-S5 ), we identified 73 putative BAP1-regulated genes ( Supplementary Tables S5F, G ). Such genes included the regulators of B cell receptor signaling Dgka and Cblb, already discussed above ( Figures 6D, E ) (91–93). The genes also included Litaf that encodes a repressor of Bcl6-expression and a regulator of autophagy and apoptosis in B cell lymphoma (94, 95), as well as Igf2bp3 that encodes a reader of the N6-methyladenosine (m6A) RNA-modification, specifically induced in GC B cells and essential for normal induction of humoral immune response ( Figures 6D, E ) (96). In summary, our study identifies novel transcriptional targets of BAP1 that are essential for the progression of GC-reaction and can contribute to the failure of humoral immunity with the loss of BAP1 function.