Pharmacologic inhibition of BAF chromatin remodeling complexes as a therapeutic approach to transcription factor-dependent cancers

Foghorn Therapeutics, 500 Technology Square, Suite 700, Cambridge, MA 02139

https://doi.org/10.7554/eLife.93478.1

Open access
Copyright information

Figures and data

| Identification and optimization of a novel class of dual inhibitors of SMARCA4/SMARCA2 ATPase activity.
(a) Chemical structures of FHT-185, FHT-1015, and FHT-2344. (b) Selective inhibition of SMARCA4 and SMARCA2 ATPases. DNA-dependent ATPase activity was measured by the ADP-glo assay in the presence of inhibitor using full length SMARCA4, SMARCA2 and CHD4. IC₅₀ values for the 3 compounds from (a) (geometric mean ± SD) are shown. (c) Representative dose response curves from (b) (mean ± SD). (d) Effects of FHT-1015 on the proliferation of cell lines were tested in a 3-day assay by Cell-titer glo (CTG). Absolute IC₅₀ values are plotted, and each dot represents an individual cell line. (e) Effects of FHT-1015 and BRM014 on the proliferation of parental MP41 or MP41-derived cells that are resistant to FHT-1015 (R1). Relative growth is plotted following a 3-day CTG assay (n=2; mean ± SD). (f-g) Inhibitory effects of FHT-1015 (f) or BRM014 (g) on wild type and mutant SMARCA4 or SMARCA2 ATPases in a Pyruvate Kinase/Lactate Dehydrogenase (PK/LDH) coupled enzyme assay (n = 3, mean ± SD).

| BAF ATPase inhibition alters enhancer accessibility of disease-relevant transcription factors.
(a) 92-1 cells were treated for 4 hours with DMSO or 100 nM FHT-1015. Heatmaps with normalized coverage tracks for DMSO or FHT-1015 treated samples are shown for all chromatin accessible regions identified in 92-1 cells by ATAC-seq. Regions were classified as consistently losing accessibility, not changing, or gaining accessibility, and sorted according to signal in the DMSO treated sample. Average signal for ATAC-seq, SMARCA4 ChIP-seq and H3K27Ac ChIP-seq was calculated and depicted as heatmaps centered at the ATAC-seq peak maximum value. Flanking region from the center of the ATAC-seq peak is 1 kb for ATAC-seq heatmap and 2.5 kb for SMARCA4/H3K27Ac ChIP-seq heatmaps. (b) Chromatin accessible regions depicted in (a) were annotated as promoter– or enhancer-associated. For each, group average signal for ATAC-seq was calculated and depicted as anchor plots centered at the ATAC-seq peak maximum value (c) Transcription factor motifs enriched at sites of lost accessibility from (a). For each region of accessible chromatin identified, potential transcription factor binding sites were assigned. The correlation between the presence of a transcription factor binding site and the observed change in chromatin accessibility upon FHT-1015 treatment was modelled using ridge regression, the coefficients of the regression were z-scored and plotted. The presence of SOX10 motifs shows the most negative coefficient indicating the stronger effect in explaining the loss of accessibility observed. (d) 92-1 super-enhancers are enriched for SOX10, MITF, TFAP2A, and SMARCA4 binding. For each enhancer/super-enhancer identified in 92-1 cells, the number of ChIP-seq peaks of the indicated factor was calculated and plotted. (e) Gene ontology (GO) of the top 500 genes mapped to 92-1 super-enhancers identified in (d) was run using PANTHER overrepresentation test. Fold enrichment for GO biological process categories with a false discovery rate (FDR) < 0.05 in a FISHER test are plotted. (f) Average ChIP-seq enrichment of SOX10, MITF, and TFAP2A at enhancers following treatment of 92-1 cells with DMSO or 100 nM FHT-1015 for 4 hours. Anchor plots are centered on ChIP-seq peaks for each indicated TF.

| BAF ATPase inhibition disrupts the SOX10-MITF transcriptional axis in uveal melanoma.
(a) Volcano plot depicting results of differential gene expression analysis of 92-1 cells treated with DMSO or FHT-1015 after 4 hours. SOX10 differential expression is highlighted in orange. Vertical dashed lines indicate 2-fold expression change thresholds; horizontal dashed line indicates adjusted p-value of 0.05. (b) Gene set enrichment analyses (GSEA) of 92-1 cells treated with DMSO or 100 nM FHT-1015 for 24h. Differential gene expression effects were ranked according to adjusted p-value and signal of change, and the ranked list was used for GSEA using MSIGDB gene sets. Enrichment of pigmentation, melanoma, and SMARCA4 target gene sets among those down-regulated by FHT-1015 is highlighted. (c) Representative GSEA plots of gene sets that are enriched among genes down-regulated by FHT-1015 in 92-1 cells. (d) Panther gene ontology enrichment of melanoma/pigmentation pathway for 53 genes characterized by loss of accessibility after a 4h treatment with FHT-1015 (ATAC-seq log₂FC < –1), loss of expression after 24h treatment with FHT-1015 (RNA-seq log₂FC < –3), and occupancy of SMARCA4 and H3K27ac by ChIP-seq in 92-1 cells (e) Heatmap representing SOX10/MITF target genes that are consistently modulated by FHT-1015 across 3 uveal melanoma cell lines. Cells were treated with DMSO or 100 nM FHT-1015 for 4h. (f) Genome browser view of RNA-seq, ATAC-seq, and ChIP-seq of SMARCA4, H3K27ac, and transcription factors at the SOX10 locus in 92-1 cells treated for 4 hours with DMSO or 100 nM FHT-1015. (g-h) PRO-seq tracks at the SOX10 gene body (g) or enhancer (h) in MP46 cells following a 1 hour treatment of DMSO or 100 nM FHT-1015. (i) 92-1 cells were treated with DMSO or 100 nM FHT-1015 for 4h or 24h and gene expression changes were plotted against the DEMETER dependency score⁴⁵. SOX10 is highlighted as a strong dependency with rapid and robust suppression of gene expression upon FHT-1015 treatment. (j) 92-1 cells were transduced with lentivirus with constitutive expression of empty vector or SOX10-Myc-DDK, and Western blot showing SOX10 protein expression after 7 days is shown alongside that of the parental 92-1 cells. (k) Resistance to FHT-1015-induced growth inhibition in 92-1 cells over-expressing SOX10 in a 3-day CTG assay. (l) 92-1 cells over-expressing SOX10 remain sensitive to IDE-196 in a 3-day CTG assay.

| SMARCA4/SMARCA2 ATPase inhibition causes uveal melanoma tumor regression.
(a) 92-1 cells were xenografted into nude mice and the effects of daily oral dosing of vehicle or FHT-2344 are plotted (n=10, mean ± SEM). (b) Representative images of tumors from (a) after 21 days of treatment. (c) Percent body weight change from (a) (n=10, mean ± SEM). (d-e) Pharmacokinetic and pharmacodynamic response of FHT-2344 in nude mice harboring 92-1 tumor xenograft. Following two days of once-daily, oral dosing, FHT-2344 was measured in the plasma (red dots, right axis), and relative gene expression levels in the tumor (bars, left axis) were analyzed by qRT-PCR for SOX10 (d) or for FGF9 (e).

| SMARCA4/SMARCA2 ATPase inhibition abrogates accessibility at master TF binding sites.
(a) Global effects on chromatin accessibility in cancer cell lines treated with FHT-1015 for 4 hours. For each cell line, the percentage of regions with differential accessibility (log₂FC ≥ 0.5, adjusted p-value < 0.05) is plotted. (b) Region-gene association plots showing the distribution of ATAC-seq Cluster 1 (loss of accessibility) peaks in relation to the TSS. For each cell line, Genomic Regions Enrichment of Annotations Tool (GREAT) algorithm was used to calculate the distance of peaks from the TSS. Number of region-gene associations (0-5 kb, 5-50 kb, 50-500 kb or >500 kb upstream of downstream of the TSS) are found on top of the corresponding bar in the graph. c) FHT-1015 reduces chromatin accessibility at lineage transcription factor binding motifs. For each cell line the enrichment of transcription factor binding motif was calculated for regions with loss of chromatin accessibility. The top 3 TF consensus binding motifs most affected and the corresponding enrichment values are indicated. (d) For each cell line the CRISPR dependency score from DepMap or Sanger Center studies was z-scored after removal of pan-essential genes. Kernel density estimation plots of all gene dependencies for each cell line are depicted on the right panel with the indicated transcription factor position in the distribution highlighted in red. (e) Cells were treated for 7 days with increasing concentrations of FHT-1015 and relative growth was measured by the CTG assay. At day 3, cells were split and fresh media and compound were replenished (n=3; mean ± SD). (f) Model summarizing the molecular mechanism of SMARCA4/SMARCA2 ATPase inhibition in TF-driven cancers (e.g., UM). (Top) in UM cells, BAF ATPase activity maintains chromatin accessibility at enhancers of the SOX10 locus, as well as at target genes, allowing for chromatin occupancy of TFs such as SOX10, MITF, and TFAP2A. This subsequently promotes expression of SOX10 and downstream transcriptional targets of these transcription factors. (Bottom) Upon BAF ATPase inhibition, chromatin is closed at these important regulatory elements, the chromatin binding of critical TFs is reduced, and gene expression is suppressed.

| SMARCA4/SMARCA2 ATPase inhibition abrogates accessibility at master TF binding sites.
(a) Global effects on chromatin accessibility in cancer cell lines treated with FHT-1015 for 4 hours. For each cell line, the percentage of regions with differential accessibility (log₂FC ≥ 0.5, adjusted p-value < 0.05) is plotted. (b) Region-gene association plots showing the distribution of ATAC-seq Cluster 1 (loss of accessibility) peaks in relation to the TSS. For each cell line, Genomic Regions Enrichment of Annotations Tool (GREAT) algorithm was used to calculate the distance of peaks from the TSS. Number of region-gene associations (0-5 kb, 5-50 kb, 50-500 kb or >500 kb upstream of downstream of the TSS) are found on top of the corresponding bar in the graph. c) FHT-1015 reduces chromatin accessibility at lineage transcription factor binding motifs. For each cell line the enrichment of transcription factor binding motif was calculated for regions with loss of chromatin accessibility. The top 3 TF consensus binding motifs most affected and the corresponding enrichment values are indicated. (d) For each cell line the CRISPR dependency score from DepMap or Sanger Center studies was z-scored after removal of pan-essential genes. Kernel density estimation plots of all gene dependencies for each cell line are depicted on the right panel with the indicated transcription factor position in the distribution highlighted in red. (e) Cells were treated for 7 days with increasing concentrations of FHT-1015 and relative growth was measured by the CTG assay. At day 3, cells were split and fresh media and compound were replenished (n=3; mean ± SD). (f) Model summarizing the molecular mechanism of SMARCA4/SMARCA2 ATPase inhibition in TF-driven cancers (e.g., UM). (Top) in UM cells, BAF ATPase activity maintains chromatin accessibility at enhancers of the SOX10 locus, as well as at target genes, allowing for chromatin occupancy of TFs such as SOX10, MITF, and TFAP2A. This subsequently promotes expression of SOX10 and downstream transcriptional targets of these transcription factors. (Bottom) Upon BAF ATPase inhibition, chromatin is closed at these important regulatory elements, the chromatin binding of critical TFs is reduced, and gene expression is suppressed.

Sign up for email alerts