Targeting SWI/SNF ATPases reduces neuroblastoma cell plasticity

Abstract

Tumor cell heterogeneity defines therapy responsiveness in neuroblastoma (NB), a cancer derived from neural crest cells. NB consists of two primary subtypes: adrenergic and mesenchymal. Adrenergic traits predominate in NB tumors, while mesenchymal features becomes enriched post-chemotherapy or after relapse. The interconversion between these subtypes contributes to NB lineage plasticity, but the underlying mechanisms driving this phenotypic switching remain unclear. Here, we demonstrate that SWI/SNF chromatin remodeling complex ATPases are essential in establishing an mesenchymal gene-permissive chromatin state in adrenergic-type NB, facilitating lineage plasticity. Targeting SWI/SNF ATPases with SMARCA2/4 dual degraders effectively inhibits NB cell proliferation, invasion, and notably, cellular plasticity, thereby preventing chemotherapy resistance. Mechanistically, depletion of SWI/SNF ATPases compacts cis-regulatory elements, diminishes enhancer activity, and displaces core transcription factors (MYCN, HAND2, PHOX2B, and GATA3) from DNA, thereby suppressing transcriptional programs associated with plasticity. These findings underscore the pivotal role of SWI/SNF ATPases in driving intrinsic plasticity and therapy resistance in neuroblastoma, highlighting an epigenetic target for combinational treatments in this cancer.

Keywords: Cancer Cell Plasticity; Core Transcription Factors; Epigenetic Plasticity; Neuroblastoma; SWI/SNF Complexes.

Figure 1. Targeting SWI/SNF ATPases inhibits NB growth.

(A, B) Comparison of SMARCA4 and SMARCA2 mRNA levels between NB (n = 33) and other cancer cell lines (n = 1446) (queried from depmap.org) (left panels). Determine SMARCA4 and SMARCA2 dependencies in NB (n = 36) and other cancer cell lines (n = 1064) by analyzing data from genome-wide cancer genetic vulnerability CRISPR screens on DepMap (right panels). (C, D) Protein levels of SMARCA2 and SMARCA4 in IMR32 cells were detected by western blot assay after treatment with different doses of PROTAC degraders ACBI1 and AU-15330. GAPDH is used as a loading control. (E, F) CellTiter-Glo assay measuring the of ACBI1 or AU-15330 treatment on cell growth (n = 3/group; Error bars indicate SEM). The cell viabilities of DMSO-treated cells are set to 100%, and IC-50 curves are generated using the GraphPad Prism software. Data are representative of two independent experiments. (G) Impact of ACBI1 or AU-15330 treatment on NB cell single spheroid growth in liquid 3D culture evaluated through Incucyte Spheroid Growth Assays (n = 4 or 6/group; Error bars indicate SEM). Data are representative of three independent experiments. Data information: In panels (A, B), data are presented as violin plots, where the middle solid lines indicate medians and the dash lines represent the 25th and 75th percentiles. The number of samples is shown on the graph. Statistical differences were calculated using a two-sided unpaired Student’s t-test. Source data are available online for this figure.

Figure 2. Depletion of SWI/SNF ATPases decreases chromatin accessibility and disrupts core TFs from binding to DNA.

(A) Analysis of differential chromatin accessibility reveals that depletion of both SMARCA2 and SMARCA4, or SMARCA4 alone, leads to reduced chromatin accessibility, as indicated by decreased ATAC-seq signals in the heatmaps. (B) HOMER de novo motif scan of sites with reduced chromatin accessibility after ACBI treatment in IMR32 cells. (C) GREAT GO analysis of ATAC-seq peaks with reduced signal intensities. (D) ATAC-seq peak distribution analysis reveals that sites with reduced chromatin accessibility are enriched in the distal regulatory regions. (E, F) Metagene plots illustrate the decreased average ChIP-seq signals in HAND2, PHOX2B, GATA3, MYCN, and H3K27ac at the sites with reduced chromatin accessibility after ACBI1 treatment or genetic silencing of SMARCA4. (G) Gene set enrichment analysis (GSEA) of RNAs-seq data shows negative enrichment of genes highly expressed in ADRN-type NB following treatment with ACBI1 for 24 h, or genetic silencing of SMARCA4 for 48 h using two different siRNAs in IMR5 cells. (H) GSEA shows negative enrichment of genes highly expressed in ADRN-type NB following treatment with ACBI1 for 24 h in IMR32 cells. Data information: In panel (B), HOMER employs the Hypergeometric Test to determine the statistical significance of the overlap between a set of observed sequences (e.g., DNA sequences containing motifs) and a set of background sequences. In panels (G) and (H), GSEA uses permutation testing to estimate the significance of the enrichment score.

Figure 3. SWI/SNF ATPases drive invasive transcriptional program and depletion of SWI/SNF ATPases suppresses NB invasion.

(A) GSEA gene ontologies (GO) analysis reveals negative enrichment of gene sets related to collagen-containing extracellular matrix and collagen fibril organization after 8 h ACBI1 treatment of IMR5 cells. (B, C) IncuCyte scratch wound healing assay demonstrates a significant decrease in relative would density upon treatment of IMR5 and BE(2)C with ACBI1 or AU-15330 (n = 4/group; Error bars indicate SEM). Data are representative of two independent experiments. (D–F) IncuCyte single spheroid matrigel invasion assay shows a significant decrease of spheroid size and invasion upon treatment of IMR5-luc-GFP, IMR32-luc-GFP and BE(2)C-luc-GFP cells with ACBI1 or AU-15330, as indicated by both bar graphs and images (n = 4 or 6/group; Error bars indicate SEM). Data are representative of three independent experiments. Data information: In panel (A), GSEA uses permutation testing to estimate the significance of the enrichment score. Source data are available online for this figure.

Figure 4. SWI/SNF ATPases maintain permissive chromatin state for MES genes in ADRN-type NB.

(A) Heatmaps display representative genes that are significantly down-regulated (p < 0.05) after genetic silencing of SMARCA4 using two different siRNAs or 8 h or 24 h treatment of IMR5 cells with ABCI1. (B) Waddington landscape depicting the plasticity of NB, where the heterogenous NB shown in the cartoon could represent a transitional NB cell population. (Created with BioRender.com). (C) RNA-seq read counts reveal significantly lower basal mRNA levels of MES genes (n = 397; Error bars indicate SEM) compared to ADRN genes (n = 345; Error bars indicate SEM), but MES gene expression is detectable in the ADRN-type NB cell line IMR32. (D) ATAC-seq read counts demonstrate similar chromatin accessibility of MES genes (n = 6112 peaks of 373 genes; Error bars indicate SEM) compared to ADRN genes (n = 5385 peaks of 276 genes; Error bars indicate SEM), while H3K27ac ChIP-seq signals on MES genes (n = 726 peaks of 276 genes; Error bars indicate SEM) are significantly lower than on ADRN genes (n = 1103 peaks of 291 genes; Error bars indicate SEM) in IMR32 cells. (E) ATAC-seq read counts demonstrate similar chromatin accessibility of MES genes (n = 6065 peaks of 382 genes; Error bars indicate SEM) compared to ADRN genes (n = 6672 peaks of 287 genes; Error bars indicate SEM), while H3K27ac ChIP-seq signals on MES genes (n = 1468 peaks of 348 genes; Error bars indicate SEM) are significantly lower than on ADRN genes (n = 2571 peaks of 331 genes; Error bars indicate SEM) in BE(2)C cells. (F) ATAC-seq read counts demonstrate similar chromatin accessibility of MES genes (n = 7957 peaks of 460 genes; Error bars indicate SE) compared to ADRN genes (n = 6211 peaks of 349 genes; Error bars indicate SEM), while H3K27ac ChIP-seq signals on MES genes (n = 1453 peaks of 355 genes; Error bars indicate SEM) are significantly lower than on ADRN genes (n = 2511 peaks of 338 genes; Error bars indicate SEM) in KELLY cells. (G, H) Analysis of ATAC-seq data reveals a significant decrease in chromatin accessibility of ADRN (n = 1779; Error bars indicate SEM), MES (n = 1508; Error bars indicate SEM), EMT (n = 601; Error bars indicate SEM), and collagen-containing extracellular matrix (CCEM) genes (n = 961; Error bars indicate SEM) that are down-regulated after ACBI1 treatment of IMR32 cells. (I) Representative MES genes that are down-regulated after SWI/SNF ATPases depletion are also found to be down-regulated after the silencing of HAND2 or MYCN in IMR32 cells from three biological replicates (n = 3; Error bars indicate SEM). Data information: In panels (C–H), data are presented as box plots, where the middle solid lines indicate the mean, and the whiskers represent min to max. In panels (C–F) and (I) the data are represented as mean ± SEM. Statistical differences were calculated using a two-sided unpaired Student’s t-test, where “ns” represents not significant. In panels (G, H), the data represent mean ± SEM, and statistical differences were calculated using a two-sided paired Student’s t-test.

Figure 5. SWI/SNF ATPases depletion reduces NB cell heterogeneity and plasticity.

(A) Representative images of SJNBL012407_X1 PDX cells cultured in neural stem-cell culture media (SCM) or shifted to culture in complete media with 10% fetal bovine serum (FBS), and after 16 h the cells were treated with DMSO (vehicle control) or ACBI1 for 4 days. Cell images reveal a distinct enlargement and flattening of mesenchymal-like morphology monolayer cells in FBS-containing media, DMSO-treated SJNBL012407_X1 cells (red arrows indicate the expanded area and adjacent cells). In contrast, ACBI1-treated cells (red arrow indicates MES-morphology type cells) predominantly exhibit spheres and neuroblast-like cells, with the rare presence of flattened, enlarged MES-morphology monolayer cells. The yellow arrow points to the semi-attached sphere. The blue arrow points to the representative neuroblast-like cell. (B) Incucyte cell confluence assays show that ACBI1 treatment of SJNBL012407_X1 reduces cell proliferation (% confluence) (n = 3; Error bars indicate SEM). Data are representative of three independent experiments. (C) Single-cell RNA-seq analysis revealed six clusters for SJNBL012407_X1 cells cultured in SCM and thirteen clusters for cells cultured in FBS-containing media. (D, E) UMAP plots show ADRN and MES signature scores in SJNBL012407_X1 cells under indicated culture conditions. Left panels: Gene signature score high cells are indicated by green and yellow dots, and circles highlight dominant cell clusters with high gene signature scores observed in at least one cell culture condition; Right panel: statistical analysis of the average ADRN (cell number: SCM, n = 790; DMSO, n = 3307; ACBI1, n = 2467) or MES (cell number: SCM, n = 880; DMSO, n = 813; ACBI1, n = 876) signature score per cell under different conditions. Data information: In the right panels of (D, E), data are presented as violin plots, where the dashed lines indicate the median and the 25th and 75th percentiles. Statistical differences were calculated using a two-sided unpaired Student’s t-test. Source data are available online for this figure.

Figure 6. Sensitization of NB cells to chemotherapeutic drug treatment through depletion of SWI/SNF ATPases.

(A) Heatmap displays the percentage of cell viability following varying doses of ACBI1 and etoposide treatment in SJNBL012407_X1 PDX cells (n = 3). The cells are cultured in FBS-containing media and initially subjected to either ACBI1 treatment alone or no treatment for 4 days. Subsequently, they are exposed to etoposide with or without prior ACBI1 treatment for an additional 3 days. Cell viabilities are measured by CellTiter-Glo Cell Viability Assay. (B) SynergyFinder online tool is used for bliss synergistic analysis to evaluate the synergistic effect of ACBI1 and etoposide treatment in SJNBL012407_X1 cells (n = 3). (C) Incucyte cell confluence assays to measure the effect of 250 nM ACBI1 or 500 nM etoposide, or ACBI1 plus etoposide treatment on cell proliferation (% confluence) (n = 3 wells/group; Error bars indicate SEM). (D) CellTiter-Glo assay measures the effect of 250 nM ACBI1 or 500 nM etoposide, or ACBI1 plus etoposide treatment on cell growth (n = 3 wells/group; Error bars indicate SEM). Cell viabilities of DMSO-treated cells are set to 100%, and the bar graph is generated using GraphPad Prism software. (E) Representative images of SJNBL012407_X1 PDX cells treated with DMSO (a vehicle control), 250 nM ACBI1, 500 nM etoposide, and ACBI1 + etoposide are presented. These images are captured using the IncuCyte SX5 imaging system. Cell images reveal distinct changes, with DMSO-treated cells showing spheres, as well as enlargement and flattening of MES-type monolayer cells. ACBI1-treated cells predominantly exhibit spheres and neuroblast-like cells, with a rare presence of flattened, enlarged MES-type cells. Etoposide-treated cells exhibit a reduced number of ADRN-type cells, but a substantial population of MES-type cells persists in the cultures. In contrast, cells treated with a combination of ACBI1 and etoposide exhibit predominantly unhealthy or decreased cells, indicated by the altered cell morphology. The yellow arrow points to the sphere, the red arrow points to the representative flattened, enlarged MES-type cell, the blue arrow points to the representative neuroblast-like cell, and the cyan arrow points to the unhealthy or dead cells indicated by the altered cell morphology, including cellular rounding, shrinkage, and detachment. (F) Heatmaps show the percentage of cell viability after different doses of ACBI1 and etoposide treatment in a panel of NB cell lines (n = 5). SK-N-SH cells are treated with the drugs for 3 days and cell proliferation is measured by Incucyte cell confluence assay. The remaining cell lines are treated with the drugs for 4 days, and cell viabilities are measured by CellTiter-Glo Cell Viability Assay. (G) SynergyFinder online tool is used for bliss synergistic analysis to evaluate the synergistic effect of the combination treatment in NB cell lines (n = 5). (H) Incucyte cell confluence assays show that the combination of ACBI1 and etoposide (Etop) treatment is more effective in suppressing NB cell proliferation (% confluency) than single agent treatment over time (n = 5 wells/group; Error bars indicate SEM). The red arrow is the time point of start to add compounds. Data information: In panel (C) the last time point and panel (D), the data are represented as mean ± SEM. Statistical differences were calculated using ordinary one-way ANOVA. All experiments (A–H) were run at least two independent times, and a representative set is shown. Source data are available online for this figure.

Figure 7. Schematic diagram illustrating the impact of SWI/SNF ATPases on ADRN-type NB cells.

High SWI/SNF ATPases levels increase chromatin accessibility, enhance DNA binding of core TFs, reduce the epigenetic barrier, and promote NB cell plasticity, contributing to intra-tumor heterogeneity, highlighting their potential as appealing therapeutic targets (Created with BioRendere.com). SE super-enhancer, TE typical enhancer.