Selective sensitivity of Ph-like B-cell acute lymphoblastic leukemia to BRG1 inhibition identifies a therapeutic vulnerability

The Philadelphia chromosome–like (Ph-like) subtype of B-cell acute lymphoblastic leukemia (B-ALL) is characterized by high relapse rates and a 3-year survival that plateaus at ~65%, necessitating the identification of novel therapeutic vulnerabilities. BRG1 ( SMARCA4 ) is an ATPase subunit of the SWI/SNF complex and plays a critical role in early B-cell development [ 1 , 2 ]. In various solid tumors, SMARCA4 is frequently mutated or deleted, acting as a tumor suppressor where its loss creates a synthetic lethal dependency on its paralog protein, BRM ( SMARCA2 ) [ 3 , 4 ]. In contrast, in hematologic malignancies such as acute myeloid leukemia (AML) [ 2 ], BRG1 is required for leukemia maintenance and enhancer-mediated MYC regulation regardless of BRM status [ 5 ]. In fact, this pharmacologic dependency on BRG1 over BRM has been extended to lymphoma [ 6 ], adrenal cancer [ 7 ], neuroblastoma [ 8 ], and rhabdomyosarcoma [ 9 ]. Here, we sought to explore BRG1’s previously uncharacterized role in high-risk subtypes of B-ALL with a special focus on therapeutic targeting. To evaluate BRG1’s role across B-cell maturation, we performed a meta-analysis of Smarca4 expression using publicly available datasets (GSE38463, GSE26408). In murine bone marrow subsets, Smarca4 mRNA was significantly elevated in Hardy fractions B, C, and D that corresponds to early pro-, late pro-, and small pre-B cells, respectively (Fig. 1a ). Analysis of a 2 nd dataset validated that Smarca4 expression is highest during early B-cell developmental stages, but not in mature B cells or mature B-cell-derived lymphomas (Fig. 1b ). Given that B-ALL originates from these early progenitors, we hypothesized that leukemic cells retain this elevated dependency for survival. Supporting this, analysis of human pediatric samples from the gene expression dataset (GSE33315) showed that SMARCA4 expression in B-ALL is generally higher than in normal CD34⁺ progenitors or T-ALL samples (Fig. 1c ). Fig. 1: B-cell acute lymphoblastic leukemia (B-ALL) depends on intact BRG1, and Ph-like B-ALL cell lines are sensitive to BRG1 inhibition in vitro. Full size image a Smarca4 expressions in different murine hardy B fractions – "CLP’: Common lymphoid progenitor, "Fr. A": Pre-Pro B, "Fr. B": Early Pro B, "Fr. C": Late Pro B / Large Pre-B, "Fr. D": Small Pre-B, "Fr. E": Immature B, and "Fr. F": Mature B cells (GSE38463). b Expression of Smarca4 in different murine B developmental stages and lymphoma models (GSE26408). c SMARCA4 expressions in B-ALL populations compared to T-Cell ALL (T-ALL), CD19 + , and CD34+ populations (GSE33315). d Western blot analysis of BRG1, BRM, and β-actin in four KMT2A -R (SEM, HB11;19, RS4;11, KOPN8) and two Ph-like (MUTZ5, MHH-CALL4) B-ALL cell lines. e Mutational frequencies of selected genes across B-ALL patients (n = 1717) from the St. Jude pediatric ALL genomic landscape study [11]. f Dependency analysis of SMARCA4 via the Cancer Dependency MAP website. Kaplan-Meier analysis of the overall survival of Ph-like B-ALL ( g ), and KMT2A -R B-ALL ( h ) patients from the COG study P9906 with high vs low SMARCA4 mRNA expression (GSE11877). i Ph-like (red; MUTZ5, MHH-CALL4) and KMT2A-R B-ALL cell lines (blue; HB11; 19, RS4;11) were treated with increasing concentrations of BRM014 (left) or FHD-286 (right). Viability/cell proliferation was determined via an XTT assay after 72 h. j GSEA enrichment plots of Ph-like (MUTZ5 and MHH-CALL4) and KMT2A-R B-ALL cell lines (HB11;19, RS4;11) after 24 h of treatment either with control or 100 nM FHD-286. “NES” refers to the Normalized enrichment score (GSE297161). k RT-PCR was performed to determine the gene expression levels of the indicated genes in Ph-like (MUTZ5, MHH-CALL4) and KMT2A- R B-ALL cell lines (HB11;19, KOPN8) after 24 h of treatment either with control or 100 nM FHD-286. ACTB (Beta Actin) was used as a reference gene to calculate expression levels. Significance was calculated via a log-rank test in Prism ( g, h ). Data are represented as individual values with mean ± SEM bars. * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001 by One-way Anova Dunnett"s multiple comparison test ( a, b ) or Two-way Anova Sidak"s multiple comparison test ( c, k ) or two-sided Log-rank (Mantel-Cox) test ( g, h ). To determine whether specific subtypes of B-ALL exhibit differential SMARCA4 expression, we next analyzed gene expression data from the COG P9906 study [ 10 ] which includes molecular-defined subsets of pediatric B-ALL. Our analysis revealed that SMARCA4 expression was highest in TCF3::PBX1 and Ph-like subtypes compared to other subtypes such as ETV6::RUNX1, Trisomy 4/10, and KMT2A -R B-ALL (Fig. S1a ). Given the relatively favorable prognosis associated with ETV6::RUNX1, Trisomy 4/10, and TCF3::PBX1, compared to the Ph-like and KMT2A -R B-ALL subtypes, we focused our subsequent analyses on the relative role(s) of BRG1 by comparing these two high-risk groups. Via western blotting and RT-PCR, we confirmed that BRG1 protein and SMARCA4 mRNA levels are higher in Ph-like B-ALL cell lines (MUTZ5, MHH-CALL4) compared to KMT2A -R B-ALL cell lines (SEM, HB11;19, RS4;11, KOPN8) (Figs. 1d and S1b ). In contrast, BRM levels were notably lower in Ph-like but higher in KMT2A -R B-ALL cell lines (Fig. 1d ). These findings were further assessed at the transcriptional level using RT-PCR, which showed similar SMARCA2 mRNA levels across both subtypes, with no consistent difference. (Fig. S1c ). Analysis of the COG P9906 gene expression dataset [ 10 ] supported these observations, and among the different B-ALL subtypes, there were no significant differences in SMARCA2 expression (Fig. S1d ). Taken together, these data suggest that the elevated BRM protein levels observed in KMT2A -R B-ALL may be regulated via post-translational modifications. Given the expression patterns of BRG1 and BRM in these subtypes, we hypothesize that Ph-like B-ALL primarily relies on BRG1, while KMT2A -R B-ALL depends on both BRM and BRG1 for chromatin remodeling. To assess the mutation frequency of SMARCA4 in leukemia, and to determine whether increased mutation rates in specific subsets of ALL might explain the lower BRG1 levels observed in KMT2A -R ALL (Fig. 1e ), we analyzed whole-genome sequencing (WGS) and whole-exome sequencing (WES) data from 1717 B-ALL patients included in the St. Jude study [ 11 ]. However, our analysis revealed that non-synonymous SMARCA4 mutations are exceedingly rare in B-ALL overall, occurring in only 0.35% of cases ( n = 6; Fig. 1e ; Table S1 ) compared to 2.6% overall mutation frequency in hematological malignancies [ 12 ] (Table S2 ), suggesting that intact BRG1 may be essential for B-ALL pathogenesis. To further characterize this dependency, we interrogated CRISPR-mediated SMARCA4 functional genomic data from the Cancer Dependency Map (DepMap) portal. While SMARCA4 has been established as a critical dependency in AML [ 2 , 5 , 13 ] and various lymphomas [ 6 ], our analysis reveals that the magnitude of this dependency is particularly striking in B-ALL (Figs. 1f and S1e ). This heightened sensitivity suggests that early B-cell states may be particularly vulnerable to BRG1 loss. Notably, B-ALL dependency on the paralog SMARCA2 was comparable to other hematologic malignancies and NSCLC (Fig. S1f, g ). This high-intensity reliance on BRG1, coupled with a lack of compensatory BRM dependency, further underscores the selective requirement for BRG1-driven chromatin remodeling in B-ALL pathogenesis. To evaluate the clinical relevance of BRG1, we stratified B-ALL patient outcomes from the COG P9906 study by median SMARCA4 expression. While survival did not differ across the entire cohort (Fig. S1h ), distinct subtype-specific patterns emerged. In Ph-like ALL, high SMARCA4 expression significantly correlated with inferior outcomes (Fig. 1g ), whereas the opposite trend was observed in KMT2A -R cases (Fig. 1h ). Conversely, elevated SMARCA2 expression was consistently associated with poor survival across all subtypes (Fig. S1i–k ). These findings suggest that Ph-like B-ALL may be particularly sensitive to BRG1-targeted strategies, while KMT2A -R cases may exhibit a divergent response. To evaluate the therapeutic potential of BRG1/BRM inhibition in B-ALL, we initially performed XTT cytotoxicity assays on two Ph-like and two KMT2A -R B-ALL cell lines in vitro, utilizing two previously tested [ 8 ] dual BRG1/BRM ATPase inhibitors - BRM014 and FHD-286. It should be noted that although currently there are no BRG1-specific inhibitors, efforts have been made to specifically target BRM [ 3 , 14 , 15 ]. As predicted, Ph-like cell lines (red) demonstrated high sensitivity to both BRM014 and FHD-286, with IC 50 values of ~50 nM and 1 nM, respectively, while KMT2A -R cell lines (blue) exhibited relative resistance to both compounds (Fig. 1i ). To understand the mechanism underlying BRG1/BRM inhibition, we performed an RNA-seq experiment to identify differentially expressed genes (DEGs) in control vs FHD-286-treated Ph-like and KMT2A -R B-ALL cell lines. Gene Set Enrichment Analysis (GSEA) revealed that genes involved in cell proliferation are strongly enriched in control samples relative to FHD-286-treated cells (Fig. 1j ). This enrichment, measured by NES (Normalized enrichment score), was more pronounced in Ph-like B-ALL cell lines compared to KMT2A -R B-ALL cell lines (Fig. 1j ). We further validated these findings via RT-PCR, confirming that key regulators of cell cycle progression, including CCND3 , CDK4 , CDK6 , E2F1 , and MYC , were downregulated following BRG1 inhibition, while the cell cycle inhibitor CDKN1B (p27) was upregulated (Fig. 1k ), consistent with a previous report showing CCND1 repression and subsequent G1 arrest in different cancers [ 8 ]. Western blot analysis corroborated MYC protein downregulation in both Ph-like and KMT2A -R B-ALL cell lines treated with BRM014 or FHD-286, consistent with previous reports that used genetic approaches to delete SMARCA4 expression [ 1 , 5 ] (Fig. S2a ). To determine whether this dysregulation of cell cycle genes causes cell cycle arrest, we performed flow cytometry-based cell cycle analysis. Treatment with BRM014 or FHD-286 resulted in a marked reduction of S-phase cells and accumulation in the G0/G1 phase across all cell lines. Ph-like B-ALL cells exhibited a nearly complete loss of S-phase cells, indicating a profound cell cycle arrest. In contrast, KMT2A -R B-ALL cells retained a substantial fraction of cells in S-phase, suggesting continued proliferation despite treatment (Fig. 2a ). Among the inhibitors, FHD-286 demonstrated relatively higher potency in inducing this arrest. Fig. 2: BRG1 inhibition results in cell cycle arrest in Ph-like B-ALL and extends the survival of Ph-like B-ALL model mice. Full size image a Cell cycle analysis of the Ph-like (MUTZ5, MHH-CALL4) and KMT2A- R B-ALL cell lines (HB11;19, KOPN8) after treatment with 100 nM BRM014 or FHD-286 for 72 h. On the left are representative examples of the flow cytometric analysis ( n = 3), and on the right is the summary of all three experiments. Kaplan-Meier survival analysis of Ph-like CDX model MUTZ5 mice ( b ) and PDX ALL2128 mice ( c ) treated with vehicle control or FHD-286 (1.5 mg/kg) ( n = 7 mice/cohort). The spleens of the MUTZ5 mice ( d ) and ALL2128 mice ( e ) were isolated after sacrifice, and spleen sizes were plotted for each treatment group. Representative images of a spleen from the Vehicle (VEH) and FHD-286 (FHD) groups are shown. f Flow cytometric analysis of percentage blood replacement of ALL2128 PDX mice by leukemic cells (hCD19+/hCD45 + ) after 35 days of treatment with vehicle or FHD-286. Data are represented as individual values with mean ± SEM bars ( d, e ). * P = 0.0333; *** P = 0.0002 by Log-rank (Mantel-Cox) test ( b, c ) or * P = 0.0333, ** P = 0.0022 by Two-tailed paired t-test ( d–f ). To assess whether BRG1/BRM inhibition induces apoptosis, we performed flow cytometry–based apoptosis assays. Neither Ph-like nor KMT2A -R B-ALL cells showed substantial apoptosis following treatment (Fig. S2b ). Consistent with this, apoptosis protein array analysis in MUTZ5 cells revealed increased anti-apoptotic proteins, including Survivin and XIAP, without activation of caspase-3 or p53 (Fig. S2c ). These results indicate that BRG1/BRM inhibition primarily suppresses leukemic growth through cell cycle arrest rather than apoptosis. To evaluate the in vivo efficacy of BRG1/BRM inhibition, we engrafted NSG mice with CRLF2 -rearranged Ph-like MUTZ5 cells and PDX ALL2128 (Fig. S2d, e ). FHD-286 treatment significantly prolonged survival in both models (log-rank P = 0.0002; Fig. 2b, c ) and was well-tolerated, evidenced by stable or improved body weight relative to controls (Fig. S2f, g ). Notably, FHD-286-treated MUTZ5 mice were sacrificed due to weight loss rather than leukemic progression, suggesting an underestimation of the survival benefit. Reduced spleen length at sacrifice confirmed lower leukemic burden in MUTZ5 mice (Fig. 2d ). In ALL2128 mice, although the spleen sizes were similar at sacrifice, a significant reduction in circulating PDX blasts after 35 days of treatment (Fig. 2e, f ) confirmed lower leukemic burden. These results provide preclinical evidence positioning BRG1/BRM inhibition as a compelling therapeutic strategy for Ph-like B-ALL. Building on this in vivo efficacy, we evaluated the potential for FHD-286 to enhance standard B-ALL chemotherapies. Using dose-response matrices and ZIP-scored viability assays, we identified robust in vitro synergy between FHD-286 and both dexamethasone and vincristine (mean ZIP δ >10; peak δ 20–31). In contrast, combinations with doxorubicin or L-asparaginase yielded only modest, dose-restricted effects with focal antagonism (Fig. S3a–d ). In summary, our data establishes BRG1 as a subtype-specific, targetable dependency in Ph-like B-ALL and provides a compelling rationale to advance BRG1-directed therapies, either with front-line chemotherapy or in combination with CAR-T immunotherapy, into clinical testing for this high-risk leukemia.