BRD9 inhibition overcomes oncolytic virus therapy resistance in glioblastoma

Abstract

Long-term survival of glioblastoma multiforme (GBM) remains challenging, spurring the development of novel therapies such as oncolytic virus therapy. While oncolytic virus shows promise in clinical trials, many patients do not respond to this therapy. Here, we perform a CRISPR screening and identify the non-canonical BRG1/BRM-associated factor (ncBAF) complex as a pivotal tumor-intrinsic factor for oncolytic virotherapy resistance. Knocking out the ncBAF-specific subunit bromodomain-containing protein 9 (BRD9) markedly augments the oncolytic efficacy of oncolytic herpes simplex virus type 1 (oHSV1) and enhances antitumor immunity. Mechanistically, BRD9 binds to RELA and potentiates the expression of downstream antiviral genes. Notably, the application of BRD9 inhibitor (IBRD9) significantly enhances the oncolytic activity of oHSV1 in various GBM models. Moreover, reduced BRD9 levels strongly correlate with improved outcomes in clinical trials of oHSV1. These findings suggest that BRD9 is an attractive target for overcoming the resistance to oHSV1 in glioblastoma treatment.

Keywords: glioblastoma; oncolytic virus; therapy resistance.

Graphical abstract

Figure 1

Genome-wide CRISPR screening reveals regulators of oncolytic virus resistance (A) Cell viability of oHSV1-treated human glioblastoma cells (MGG4, GSC3264, and BNI21) and normal neural cells (hNP1, ENSA, and iPSC-derived neuron) (0.2 MOI, 24 hpi) (n = 3). (B) Schematic illustration of cerebral organoid and patient-derived glioblastoma tumor section preparation with oHSV1 treatment. Analysis of oHSV1 replication in cerebral organoids and patient-derived glioblastoma tumor sections. After infection with oHSV1 (1 × 10⁵ PFU, 24 hpi), the distribution of oHSV1 in the cerebral organoids and tumor sections was detected by anti-HSV1 immunohistochemistry. Scale bars, 100 μm. (C) Schematic illustration of the CRISPR screening for oHSV1 treatment in CT2A^Nectin1 cells. (D) Venn diagram showing the intersection of essential genes from the oHSV1 treatment vs. control comparison and control vs. start point comparison (left). Gene Ontology analysis of the oHSV1-specific candidate genes (right). (E) Scatterplot for the β scores of each gene in the oHSV1 treatment group versus the control group and control group versus the start point group. The selected genes (921) in control vs. the start point comparison were gated in red frame. The selected genes (402) in oHSV1 treatment vs. control comparison were gated in blue frame. The oHSV1-specific genes (351) were gated in black frame. The positive selection Nectin1 (black), canonical BAF subunits (green), non-canonical BAF subunits (red), and polybromo-associated BAF-specific subunits (blue) are highlighted. (F) Diagram showing three different variants of switch/sucrose non-fermentable complex (SWI/SNF) complexes: canonical BAF (cBAF), noncanonical BAF (ncBAF), and polybromo-associated BAF (PBAF). cBAF-specific subunits (Arid1a and Arid1b) are colored green, ncBAF-specific subunits (Brd9, Gltscr1, and Gltscr1l) are colored red, and PBAF-specific subunits (Arid2, Brd7, and Pbrm1) are colored blue. Shared components among the complexes are colored gray (D1: Smarcd1; D2: Smarcd2; D3: Smarcd3; E1: Smarce1; B1: Smarcb1; A2: Smacra2; A4: Smacra4; Ab: β-actin; C1: Smacrc1; C2: Smarcc2; A6a: Actl6a; B7: Bcl7). (G) Cell viability of oHSV1-treated CT2A^Nectin1 cells (0.2 and 0.5 MOI, 24 hpi) transfected with sgRNA targeting the indicated SWI/SNF complex-specific subunits (n = 3). Data represent mean ± SD. Unpaired two-tailed Student’s t test (A), one-way ANOVA (G). The diagrams (B and C) were created using BioRender. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Figure 2

Knockout of BRD9 enhances the sensitivity of glioblastoma cells to oHSV1-mediated ICD and potentiates viral replication (A) Flow cytometry analysis of oHSV1-treated control or Brd9-deficient CT2A^Nectin1 cells subjected to PI/Annexin V staining for cell death analysis (PI+) (n = 3). (B) Flow cytometry analysis of oHSV1-treated control or BRD9-deficient MGG4 cells subjected to PI/Annexin V staining for cell death analysis (PI+) (n = 3). (C) Quantitative real-time PCR analysis of oHSV1 glycoprotein D levels in oHSV1-treated control or BRD9-deficient glioma cells (CT2A^Nectin1 and MGG4). GAPDH transcript normalization (n = 3). (D) Schematic illustration of the plaque formation assay in Vero cells. (E) Plaque formation assay using culture medium from oHSV1-treated control or BRD9-deficient glioma cells (CT2A^Nectin1 and MGG4). Left: representative images of crystal violet-stained Vero cell plaques treated with different culture medium (n = 3). (F) Calreticulin (CRT) exposure analysis of oHSV1-treated control or Brd9-deficient CT2A^Nectin1 cells (n = 3). (G) Calreticulin (CRT) exposure analysis of oHSV1-treated control or BRD9-deficient MGG4 cells (n = 3). (H) Extracellular ATP level analysis in control or BRD9-deficient glioma cells (CT2A^Nectin1 and MGG4) treated with oHSV1 (n = 3). (I) Extracellular HMGB1-level analysis in control or BRD9-deficient glioma cells (CT2A^Nectin1 and MGG4) treated with oHSV1 (n = 3). Data represent mean ± SD. Two-way ANOVA (A, B, F, G, H, and I), one-way ANOVA (C and E). The diagram (D) was created using BioRender. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Figure 3

Knockout of Brd9 increases the efficacy of oHSV1 therapy in mouse glioma models and augments the corresponding immune cell response (A and B) GBM mouse models were established by intracranial injection of 2 × 10⁴ control or Brd9-deficient mouse glioma cells (CT2A^Nectin1 and GL261^Nectin1) into C57BL/6J mice. 3 days later, the mice were intratumorally injected with 6 × 10⁵ PFU of oHSV1 or mock treatment. (A) Survival curve of control or Brd9-deficient CT2A^Nectin1 tumor-bearing mice treated with oHSV1 or mock treatment (n = 6). (B) Survival curve of control or Brd9-deficient GL261^Nectin1 tumor-bearing mice treated with oHSV1 or mock treatment (n = 6). (C) Analysis of oHSV1 replication in the brains of control or Brd9-deficient CT2A^Nectin1 tumor-bearing mice. The samples were collected 2 days after intratumoral injection with 6 × 10⁵ PFU of oHSV1; the distribution of oHSV1 in the tumors was detected via anti-HSV1 immunohistochemistry. Scale bars, left, 100 μm; right, 50 μm; (n = 3). (D) Flow cytometry analysis of the percentages of CD3⁺ T cells and cDC1s (CD11c+MHCII+XCR1+) in the control or Brd9-deficient CT2A^Nectin1 tumor samples within oHSV1 treatment (n = 3 in mock group and n = 4 in oHSV1 treatment group). (E) scRNA-seq analysis of CD45⁺ cells from oHSV1-treated control or Brd9-deficient CT2A^Nectin1 tumor-bearing mice. The CD45⁺ tumor-infiltrating leukocytes were sorted with anti-mouse CD45 magnetic beads and combined in equal numbers within the same groups for 10× Genomics scRNA-seq (n = 4 mice per group). Uniform manifold approximation and projection (UMAP) plot of the scRNA-seq data depicting the different immune cell subsets (left). The proportion of cells in each cluster in the oHSV1-treated Brd9 deficiency group vs. the oHSV1-treated control group (right). Positive values indicate an increase in cluster occupancy following Brd9 deficiency. (F) UMAP plot of T cell subclusters (left). The proportion of cells in each subcluster in the oHSV1-treated Brd9-deficient group vs. the oHSV1-treated control group (right). Positive values indicate an increase in cluster occupancy following Brd9 deficiency. (G) Characterization of clusters with functional T cell markers. Dot plot shows the average expression levels and cell expression proportions of selected T cell marker genes (top); larger dots indicate a higher proportion of cells with expression, and red versus blue indicates higher expression. Heatmap shows the scaled expression of T cell functional markers (bottom). (H) Flow cytometry analysis of the percentages of CD8⁺ T cells and CD8⁺ T cells expressing functional markers (CD69 and GZMB) in the control or Brd9-deficient CT2A^Nectin1 tumor samples within oHSV1 treatment (n = 3 in mock group and n = 4 in oHSV1 treatment group). (I) Flow cytometry analysis of the percentage of central memory CD8⁺ T cells (TCM, identified as CD44⁺ and CD62L+ in CD8⁺ cells) in the control or Brd9-deficient CT2A^Nectin1 tumor samples within oHSV1 treatment (n = 4). Data represent mean ± SD. Log rank test (A and B), unpaired two-tailed Student’s t test (C), two-way ANOVA (D, H, and I). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Figure 4

BRD9 interacts with RELA to regulate antiviral gene expression (A) Volcano plot illustrating differential gene expression between oHSV1-treated control MGG4 cells and oHSV1-treated BRD9 knockout MGG4 cells. Genes with log2(fold change) > 1 and false discovery rate (FDR) < 0.05 are marked in red. Genes with log2(fold change) < −1 and FDR < 0.05 are marked in blue. Other genes are marked in gray. (B) Pathway enrichment analysis of genes downregulated in BRD9 knockout MGG4 cells following oHSV1 treatment. (C) Gene set enrichment analysis (GSEA) of TNFA signaling via NFKB pathway in oHSV1-treated BRD9 knockout MGG4 group vs. oHSV1-treated control MGG4 group. (D) Quantitative real-time PCR analysis of antiviral gene (BST2, ISG20, and OPTN) expression in control and BRD9-deficient MGG4 cells upon oHSV1 treatment. GAPDH transcript normalization (n = 3). (E) Pie chart representing BRD9 binding to genes that were significantly downregulated in BRD9 knockout cells after oHSV1 infection, as determined by BRD9 ChIP-seq. (F) Pathway enrichment analysis of BRD9 binding to genes that were significantly downregulated in BRD9 knockout cells after oHSV1 treatment. (G) Histogram representation of ChIP read density (±1 kb) at BRD9 and RELA common binding sites in control and BRD9 knockout MGG4 cells. (H) Genome browser tracks of BRD9 ChIP-seq signals and RELA ChIP-seq signals (GSM2394420) at BST2, ISG20, and OPTN loci. (I) CoIP analysis of the interaction between RELA and BRD9 in nuclear extracts from MGG4 cells. (J) Scatterplot for the β scores of each gene in the oHSV1 treatment vs. the control comparison and the control vs. start point comparison. Rela (green) and Brd9 (red) are highlighted. (K) Western blot analysis of the RELA knockout efficiency in MGG4 cells. (L) Flow cytometry analysis of cell death in oHSV1-treated control or RELA-deficient MGG4 cells, as indicated by PI/Annexin V staining (PI+) (n = 3). (M) Quantitative real-time PCR analysis of oHSV1 glycoprotein D levels in control or RELA-deficient MGG4 cells following oHSV1 treatment. GAPDH transcript normalization (n = 3). (N) Quantitative real-time PCR analysis of antiviral genes (BST2, ISG20, and OPTN) in control or RELA-deficient MGG4 cells following oHSV1 treatment. GAPDH transcript normalization (n = 3). Data represent mean ± SD. Two-way ANOVA (D, L, and N) and unpaired Student’s t test (M). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Figure 5

Pharmaceutically targeting BRD9 enhances the antitumor effect of oHSV1 in vitro (A) Flow cytometry analysis of oHSV1-treated control or IBRD9-pretreated (1 μM, 24 h) CT2A^Nectin1 and MGG4 cells subjected to PI/Annexin V staining for cell death analysis (PI+) (n = 3). (B) Quantitative real-time PCR analysis of oHSV1 glycoprotein D levels in control or IBRD9-pretreated (1 μM, 24 h) CT2A^Nectin1 and MGG4 cells treated with oHSV1. GAPDH/Gapdh transcript normalization (n = 3). (C) Plaque formation assay of oHSV1-treated control or IBRD9-pretreated (1 μM, 24 h) CT2A^Nectin1 and MGG4 cell culture medium. The virus titer was determined after 48 h (n = 3). (D) Calreticulin (CRT) exposure analysis of control or IBRD9-pretreated (1 μM, 24 h) CT2A^Nectin1 cells treated with oHSV1 (n = 3). (E) Calreticulin (CRT) exposure analysis of control or IBRD9-pretreated (1 μM, 24 h) MGG4 cells treated with oHSV1 (n = 3). (F) Extracellular ATP level analysis in control or IBRD9-pretreated (1 μM, 24 h) CT2A^Nectin1 and MGG4 cells treated with oHSV1 (n = 3). (G) Extracellular HMGB1 level analysis in control or IBRD9-pretreated (1 μM, 24 h) CT2A^Nectin1 and MGG4 cells treated with oHSV1 (n = 3). (H) In vitro co-culture proliferation experiments with OT-I CD8⁺ T cells and cDC1s generated from WT mice in the IBRD9- and oHSV1-treated CT2A^Nectin1-OVA-B2m^−/− cells (n = 3). (I) Schematic of human glioblastoma-derived organoid processing and verification of oHSV1-mediated killing by PI staining, 3D cell titer assays, and ICD marker analysis. (J) PI staining of IBRD9-pretreated (1 μM, 24 h) human glioblastoma-derived organoids treated with oHSV1. Scale bars, 300 μm. (K) 3D cell viability assay of IBRD9-pretreated (1 μM, 24 h) human glioblastoma-derived organoids treated with oHSV1 (n = 3). (L) Extracellular ATP-level analysis in control or IBRD9-pretreated (1 μM, 24 h) human glioblastoma-derived organoids treated with oHSV1 (n = 3). (M) Extracellular HMGB1-level analysis in control or IBRD9-pretreated (1 μM, 24 h) human glioblastoma-derived organoids treated with oHSV1 (n = 3). (N) Schematic of human glioblastoma-derived tumor slice processing and verification of oHSV1 replication by anti-HSV1 staining and quantitative real-time PCR. (O) Representative immunohistochemistry images of HSV1 staining in IBRD9-pretreated or control human glioblastoma-derived tumor slices. Scale bars, 50 μm. (P) Analysis of oHSV1 replication in IBRD9-pretreated (2 μM, 24 h) or control human glioblastoma-derived tumor slices. After oHSV1 treatment, the distribution of oHSV1 in the sections was detected by anti-HSV1 immunohistochemistry (n = 3). (Q) Quantitative real-time PCR analysis of oHSV1 glycoprotein D levels in control or IBRD9-pretreated (2 μM, 24 h) human glioblastoma-derived tumor sections treated with oHSV1. GAPDH transcript normalization (n = 3). Data represent mean ± SD. Two-way ANOVA (A, D, E, F, G, H, K, L, and M), unpaired two-tailed Student’s t test (B, C, P, and Q). The diagrams (I and N) were created using BioRender. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Figure 6

Pharmaceutically targeting BRD9 enhances the antitumor effect of oHSV1 in vivo, and BRD9 expression is associated with poor clinical outcome in cancer patients treated with oHSV1 (A) Survival curve of IBRD9, oHSV1, and ICB (the anti-mouse PD-1 antibody and the anti-mouse CTLA4 antibody) combination therapy in CT2A^Nectin1 tumor-bearing mice (n = 5). (B) Survival curve of IBRD9, oHSV1, and ICB (the anti-mouse PD-1 antibody and the anti-mouse CTLA4 antibody) combination therapy in GL261^Nectin1 tumor-bearing mice (n = 5). (C) Survival curve of long-term survivor mice and age-matched control mice challenged with CT2A^Nectin1 cells (n = 5). (D) Survival curve of long-term survivor mice and age-matched control mice challenged with GL261^Nectin1 cells (n = 5). (E) Kaplan-Meier analysis showing the PFS of patients with glioblastoma treated with oHSV1 subdivided by the expression of BRD9 (high-expression patients: n = 7; low-expression patients: n = 6). HR, hazard ratio. (F) Analysis of BRD9 expression in liver cancer and pancreatic cancer biopsy sections from patients participating in an oHSV1 clinical trial. Representative immunohistochemistry images of BRD9 staining (oHSV1 response: SD, stable disease; oHSV1 nonresponse: PD, progressive disease). Scale bars, 50 μm. (G) BRD9-stained sections were quantified by H-score (SD: n = 14; PD: n = 13). Data represent mean ± SD. Unpaired two-tailed Student’s t test (G) and log rank test (A–E). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.