SMARCA4 is a haploinsufficient B cell lymphoma tumor suppressor that fine-tunes centrocyte cell fate decisions

Abstract

SMARCA4 encodes one of two mutually exclusive ATPase subunits in the BRG/BRM associated factor (BAF) complex that is recruited by transcription factors (TFs) to drive chromatin accessibility and transcriptional activation. SMARCA4 is among the most recurrently mutated genes in human cancer, including ∼30% of germinal center (GC)-derived Burkitt lymphomas. In mice, GC-specific Smarca4 haploinsufficiency cooperated with MYC over-expression to drive lymphomagenesis. Furthermore, monoallelic Smarca4 deletion drove GC hyperplasia with centroblast polarization via significantly increased rates of centrocyte recycling to the dark zone. Mechanistically, Smarca4 loss reduced the activity of TFs that are activated in centrocytes to drive GC-exit, including SPI1 (PU.1), IRF family, and NF-κB. Loss of activity for these factors phenocopied aberrant BCL6 activity within murine centrocytes and human Burkitt lymphoma cells. SMARCA4 therefore facilitates chromatin accessibility for TFs that shape centrocyte trajectories, and loss of fine-control of these programs biases toward centroblast cell-fate, GC hyperplasia and lymphoma.

Keywords: B-cell; BAF; SMARCA4; SWI/SNF; epigenetics; germinal center; immunology; lymphoma; transcription.

Figure 1:. SMARCA4 selectively regulates the GC reaction.

A) Frequency of SMARCA4 inactivating mutations in lymphoma. B-C) RNA (B) and protein (C) expression of SMARCA4 and its paralog, SMARCA2, in normal B cell development. The box plot center line represents the median, boxes span the second and third quartile, and whiskers show the minimum and maximum. D) The floxed allele for Smarca4 generated for this study. E-J) The frequency of total B cells (Cd19⁺B220⁺) and developmental stages from 8-10 week-old Smarca4^+/+ and Smarca4^+/− mice (Cd19-Cre) in the bone marrow (E-F), blood (G-H), and spleen (I-J). The center line and error bars represent the mean +/− the standard error of the mean. K) Schema, example flow cytometry contour plots and frequencies of GC B cells in SRBC-immunized Smarca4^+/+ and Smarca4^+/− mice (Cd19-Cre). The center line and error bars represent the mean +/− the standard error of the mean. Comparisons use unpaired Students T-test. ns, not significant. See also Figure S1.

Figure 2:. Lymphomagenesis in mice with GC-specific Smarca4 loss combined with Myc over-expression.

A) Enlarged lymph nodes and spleen-to-body weight ratio in mice following three SRBC immunizations. The center line and error bars represent the mean +/− the standard error of the mean. P-values calculated with unpaired Student’s T-test. B) Lymph node histomorphology by H&E staining and immunohistochemistry for B220 and Ki-67 following three SRBC immunizations. C-D) Examples of histomorphology and immunohistochemistry (B220 and Ki-67) of lungs from mice following three SRBC immunizations. The scale bar in panel C represents 1mm. The center line and error bars in panel D represent the mean +/− the standard error of the mean. P-values calculated with unpaired Student’s T-test. E) Overall survival of mice with/without Smarca4 loss and presence/absence of Myc over-expression. P-value calculated with log-rank test. F) Examples of lymph node gross appearance (1cm x 1cm grid) and histology from ill mice. Boxes outline the region of higher magnification. Scale bars represent 100μM. See also Figure S2.

Figure 3:. GC hyperplasia following Smarca4 loss.

A) Schema, example flow cytometry and frequencies of GC B cells in SRBC-immunized Smarca4^+/+ and Smarca4^+/− mice (Aid-Cre). B) PCR genotyping of sorted GC B cells of Smarca4^+/+, Smarca4^+/− and Smarca4^−/− mice. C) Examples of histomorphology by H&E and immunohistochemistry (B220 and PNA) in spleens from immunized mice. D-E) Enumeration of GCs per section (D) and GC size (E) determined by PNA staining. F) Enumeration of GC B cell frequency by flow cytometry at day 5, 10, 15, 20 post-SRBC immunization. G-H) Example flow cytometry plots (G) and frequencies (H) of centroblasts and centrocytes among splenic GC B cells. The center line and error bars for panels A, D, E, F and H represent the mean +/− the standard error of the mean. All P-values were calculated with an unpaired Students T-test. ns, not significant. See also Figure S3.

Figure 4:. GC competitive advantage with Smarca4 loss.

A-C) Gating of non-GC and GC B cells (A) and Smarca4^+/+ [Cd45.1/2] and Smarca4^+/− [CD45.2] populations (B), and relative fractions of Smarca4^+/+ (black) and Smarca4^+/− (red) cells from the same mouse (C), normalized to non-GCB ratios. D-F) Centroblast [CB] and centrocyte [CC] gating (D) and the enumeration of Smarca4^+/+ and Smarca4^+/− cells within each population (E-F). G-I) EdU⁺ GC B cell gating (G) and the enumeration of Smarca4^+/+ and Smarca4^+/− cells within EdU⁺ centroblasts [CB] and centrocytes [CC] (H-I). J-L) Annexin-V⁺ (AnnV⁺) GC B cell gating (J) and the enumeration of Smarca4^+/+ and Smarca4^+/− cells within AnnV⁺ centroblasts [CB] and centrocytes [CC] (K-L). All P-values were calculated with a paired Students T-test. ns, not significant.

Figure 5:. Altered GC composition associated with Smarca4 loss.

A) UMAP of WNN clustering of scMO data from Smarca4^+/+ and Smarca4^+/− mice, colored by cell state. Individual density UMAPs (right) show cells from each genotype. B) Frequencies of cells within each cell state shown in panel A, compared with a Fisher exact test. The center line represents the median, boxes span the second to third quartile, and whiskers span from minimum to maximum. *P<0.05; **P<0.01; ***P<0.001. C-D) Gating of GFP+ recycling cells, and their enumeration by genotype in GC B cells (D) and within centroblasts and centrocytes (D). The center line and error bars for panel D represents the mean +/− the standard error of the mean. E-H) Gating (E) and enumeration of memory B cells (F) and plasma cells (G-H) from the spleens of immunized and boosted mice. The center line and error bars for panels G-H represent the mean +/− the standard error of the mean. I-L) Gating (I) and enumeration of memory B cells (J) and plasma cells (K-L) from the bone marrow of immunized and boosted mice. The center line and error bars for panels J-L represent the mean +/− the standard error of the mean. P-values in panels C-L were calculated with an unpaired Students T-test. ns, not significant. See also Figures S4 and S5.

Figure 6:. Reduced SPI1, IRF, and NFκB activity in murine centrocytes with Smarca4 loss.

A) ChromVar analysis of TF motifs with significantly reduced activity in centroblasts [CB] and centrocytes [CC] from Smarca4^+/− mice, ordered by average rank, and their activity in wild-type centroblasts and centrocytes. B) The density of GC populations from Smarca4^+/+ mice (above) and the relative difference in densities between Smarca4^+/+ and Smarca4^+/− mice (below) along the pseudotime trajectory. Lines span minimum to maximum and box represents IQR. C) Relative change in TF activity for SPI1, IRF1, IRF8 and IRF4 across pseudotime, aligned with Figure 6B. Negative values indicate reduced activity in Smarca4^+/− cells (significance of P<1x10⁻⁷ colored maroon). Scale bar represents 200μM. D) Heatmaps of regions with significantly reduced chromatin accessibility in bulk sorted centrocytes [average of 4 replicates]. Smarca4 binding is shown from a published dataset for the same regions. E) Change in TF activity in Smarca4^+/− compared to Smarca4^+/+ centrocytes determined by TOBIAS. F) Hypergeometric enrichment of gene sets among genes that are significantly down-regulated in Smarca4^+/− or Pbrm1^−/− centrocytes (CC) compared to Smarca4^+/+ or Pbrm1^+/+ centrocytes, respectively. G) Preranked GSEA of GC terminal differentiation genes in sorted centrocytes assessed by ATAC-seq. Negative enrichment scores indicate reduced accessibility in the Smarca4^+/− condition. H-I) Enrichment of predicted TF binding calculated by overlap of peaks from public datasets (H) or over-representation of TF motifs (I) in regions with increased chromatin accessibility in wild-type centrocytes compared to centroblasts (top), or decreased chromatin accessibility in Smarca4^+/− (middle) or Pbrm1^−/− (bottom) centrocytes compared to their wild-type controls. See also Figure S6 and Tables S1 and S2.

Figure 7:. Loss of SPI1 and IRF activity in human BL cells accompanied by repression of BCL6 target genes.

A) Schematic of AuD system for SMARCA4 degradation, showing loss of GFP-tagged SMARCA4 over time. B) Example gating for sorting. SMARCA4 protein expression in sorted populations is in Figure S7. C) ssGSEA for genes with reduced expression and chromatin accessibility following SMARCA4 degradation in the Raji-AuD cell line, comparing primary pediatric BL tumors with wild-type or mutant SMARCA4. P-value calculated with an unpaired Student’s T-test. D) Protein expression of SMARCA4 in CRISPR-modified Raji cell lines. E) Preranked GSEA of genes that are significantly reduced following SMARCA4 degradation in the Raji-AuD cell line, comparing isogenic SMARCA4^WT and SMARCA4^R1192C Raji (left) and CA46 (right) cell lines. F) Preranked GSEA enrichment scores for published gene sets comparing GFP-low to GFP-high Raji-AuD cells, or isogenic SMARCA4^R1192C to SMARCA4^WT cells for the Raji and CA46 cell lines. Negative enrichment scores indicate reduced expression in the SMARCA4 loss/mutant condition. G) Gene expression changes of genes from the leading edge of enriched SPI1, IRF4 and BCL6 target gene sets. Negative log2 fold-change indicates reduced expression in the SMARCA4 loss/mutant condition. H) Enrichment of regions with significantly reduced chromatin accessibility associated with SMARCA4 degradation (Raji AuD) or mutation (R1192C) in BL cell lines using overlap with peaksets from published ChIP-sequencing studies in ChIP-atlas (above), or motif enrichment using HOMER (below). I) BCL6 luciferase reporter activity in sorted GFP-high and GFP-low populations from the Raji-AuD system. Data are normalized to the 24h GFP-high condition, error bars represent the standard error of the mean. P-values were calculated with an unpaired Student’s T-test. J) Enrichment scores for preranked GSEA comparing experimental conditions (denoted by color) to the control GFP-high DMSO-treated condition. Adjusted P-values were calculated by preranked GSEA (****adj. P<0.001) and negative enrichment scores indicate reduced expression in the experimental condition compared to control. Full gene set names: FX1-induced, “BCL6_Targets_GCBDLBCLcellLine_FX1_induced”; BCL6 targets, “GCB.promoter.BCL6.SMRT.BCOR”; SPI1 targets, “PU.1_target_ChIPseq_Ebert_Cell”. See also Figure S7 and Tables S3-S5.

Figure 8:. Temporal patterns of reduced chromatin accessibility, TF activity and cooperative binding following SMARCA4 degradation.

A) TF activity determined by TOBIAS, comparing GFP-low and GFP-high cells from cultures treated with Dox + IAA for 24h, 48h or 72h. B) Heatmaps of regions with reduced accessibility acquired at early (top), intermediate (Int, middle) or late (bottom) time points. Increase line stroke indicates regions that are significantly reduced (q-value<0.05, fold-change>1.5) in GFP-low cells at a given time point. C) Enrichment of ChIP-seq-defined TF motifs in regions with early, intermediate or late loss of chromatin accessibility. D) Fold-changes in SPI1 binding in GFP-low compared to GFP-high cells (average of 3 replicates) at each time-point, showing regions with reduced chromatin accessibility that are predicted to be bound by SPI1 using TOBIAS. Negative log2 fold-change indicates reduced binding in the GFP-low population. E) Read densities for ATAC-sequencing and SPI1 CUT&RUN from GFP-high and GFP-low cells at each timepoint (average of 3 replicates) for the SPI1 target gene, HAVCR2. The region of significant chromatin accessibility loss is highlighted in yellow and SPI1 motifs and published BCL6 ChIP-sequencing data are included for reference. F) Change in accessibility between GFP-low and GFP-high cells at 24h by the distance of the nearest SPI1 and IRF1/4 motif (determined by TOBIAS) to the ATAC peak summit. Center point represents the mean and the line represents the standard error of the mean. ***Wilcoxon rank-sum test P-value<<<0.001. G) Change in accessibility of ATAC peaks across all time points demarcated by relative distances of SPI1 and IRF1/4 motifs from the ATAC peak summit. H) Significance of composite elements across early, intermediate and late losses in chromatin accessibility. See also Figure S8 and Tables S6 and S7.