Selective PROTAC-mediated degradation of SMARCA2 is efficacious in SMARCA4 mutant cancers

Abstract

The mammalian SWItch/Sucrose Non-Fermentable (SWI/SNF) helicase SMARCA4 is frequently mutated in cancer and inactivation results in a cellular dependence on its paralog, SMARCA2, thus making SMARCA2 an attractive synthetic lethal target. However, published data indicates that achieving a high degree of selective SMARCA2 inhibition is likely essential to afford an acceptable therapeutic index, and realizing this objective is challenging due to the homology with the SMARCA4 paralog. Herein we report the discovery of a potent and selective SMARCA2 proteolysis-targeting chimera molecule (PROTAC), A947. Selective SMARCA2 degradation is achieved in the absence of selective SMARCA2/4 PROTAC binding and translates to potent in vitro growth inhibition and in vivo efficacy in SMARCA4 mutant models, compared to wild type models. Global ubiquitin mapping and proteome profiling reveal no unexpected off-target degradation related to A947 treatment. Our study thus highlights the ability to transform a non-selective SMARCA2/4-binding ligand into a selective and efficacious in vivo SMARCA2-targeting PROTAC, and thereby provides a potential new therapeutic opportunity for patients whose tumors contain SMARCA4 mutations.

Fig. 1. A947 is a potent and moderately selective degrader of SMARCA2.

a Chemical structure of A947. b Dose-response curves of A947 displacing a biotinylated SMARCA2/4-binding probe from recombinant SMARCA2 and SMARCA4 bromodomains. Data are presented as mean ± s.d. from 4 replicates. c Quantification of SMARCA2 and SMARCA4 protein levels by In Cell Western^TM following 20 h treatment of SW1573 cells. Data are normalized to DMSO control-treated cultures and presented as mean ± s.d. from 7 independent experiments. d Immunoblot analysis of respective proteins following 18 h treatment of SW1573 cells with a dose-response of A947. HDAC1 serves as a loading control. Data is representative of 2 independent experiments. e Licor-based quantification of SMARCA2 and SMARCA4 isoforms ectopically expressed in TOV112D cells following 24 h treatment with A947. Data is normalized to DMSO control treated cells. UniProt identifiers of the respective isoforms are shown in parentheses. f Pretreatment (1 h) of SW1573 cells with 20-fold molar excess of the respective inhibitors can block A947 (500 nM) -mediated degradation of SMARCA2. Total ubiquitin levels serve as a control for MLN-7243 and MG-132. HDAC1 serves as a loading control. g Immunoblots demonstrating FLAG-tagged SMARCA2 ortholog expression in human TOV112D and murine LA4 cells. Tubulin serves as a loading control. h Licor-based quantification of FLAG-tagged orthologs (human, mouse, rat) of SMARCA2 ectopically expressed in human (TOV112D) and murine (LA4) cell lines upon 24 h treatment with a dose-response of A947. Data are normalized to levels of the respective SMARCA2 ortholog in control (DMSO) lysates. Data in (f,g) were confirmed in 3 similar experiments. Source data are provided as a Source Data file.

Fig. 2. A947-mediated ubiquitination and degradation of SMARCA2/4.

a Global ubiquitylome changes, as assessed by di-glycine reminant mass spectrometry profiling, in SW1573 cells following treatment with 500 nM A947 for 5 and 30 min. n = 5720 unique ubiquitinated peptides were identified and presented as log2 fold change relative to DMSO control cultures. Positions of ubiquitinated lysines in peptides that uniquely map to SMARCA2 (UniProtKB P-51531-1) or SMARCA4 (UniProtKB P-51532-2) are highlighted in blue and red, respectively. Peptides shared between SMARCA2/4 are shown in green, with ubiquitinated lysines mapped to the SMARCA2 sequence. b Global proteome assessed by mass spectrometry following 8 h treatment with A947 (100 nM) in SW1573 cells. Data are presented as a log2 fold change in the abundance of the respective proteins in two biological replicates. ~8900 and ~8400 proteins were quantified in replicate 1 and 2, respectively.

Fig. 3. A947-mediated inhibition of cell proliferation in SMARCA4-mutant NSCLC cell lines.

a Viability of NCI-H1944 following 7 days of treatment with a dose response of A947 or binding-defective epimers. Error bars represent mean ± SD from n = 3 biologic replicates. b Immunoblots of SMARCA2, SMARCA4 and PBRM1 levels across a panel of NSCLC cell lines defined by SMARCA4 gene mutation status. β-tubulin serves as a loading control. Data are representative of 2 independent experiments. c Effect of A947 treatment on the growth of 30 lung cancer cell lines defined by SMARCA4 or SMARCA2 status, represented as the concentration of A947 required to inhibit growth by 50% (IC₅₀) following 7 days of treatment. Individual cell line IC₅₀’s were determined from n = 3 biologic replicates. Median IC₅₀’s across models defined by mutational status are indicated by the black line. Significance was assessed by a two-tailed, Mann Whitney test. Asterick indicates p = 0.0003 d Cell cycle distribution following 48 h treatment of a dose response of A947 across 4 SMARCA4-mutant, 2 SMARCA4-wt and one SMARCA2-deficient NSCLC cell lines. Aphidicolin (Aph, 1 μM) treatment served as a control to block entry into S phase for the SMARCA4-wt and SMARCA2-deficient models. Error bars represent mean ± SD from n = 3 biologic replicates. e Log₂-transformed fold change in mRNA expression values, as measured by RNA-seq, in HCC2302 cells treated for 96 h with A947 compared to control DMSO treated cultures (x-axis), as well as HCC2302-shSMARCA2 cells treated with doxycycline for 168 h compared to HCC2302 cells expressing a non-targeted control shRNA (shNTC). RNAseq data is representative of triplicated cultures. The correlation was calculated by the Pearson coefficient. p = 2.2e−16. Source data are provided as a Source Data file.

Fig. 4. A947 is selectively efficacious in SMARCA4-mutant NSCLC xenograft models.

a Tumor concentration (upper graphic) and pharmacodynamic biomarker responses (lower graphic) monitored over a 2 week period in HCC515 xenograft tumors following a single-dose, intravenous (i.v.) administration of A947 (40 mg/kg). Tumor levels of SMARCA2 protein were quantified by ImageLab from Western blots and normalized to a loading control protein (β-actin). Tumor levels of the mRNA transcripts, KRT80 and PLAU, were quantified by Taqman. Data are presented relative to levels in untreated tumors and represented as mean ± s.d from n = 5 animals per timepoint. b, c Tumor volume in mice harboring either SMARCA4-mutant HCC515 (b) or HCC2302 (c) xenografts following administration of A947 (40 mg/kg, i.v.) or vehicle control. Data are presented as mean ± s.e.m. (n = 10 mice/group). Statistical significance was assessed by a two-sided, unpaired Student t test. p = 0.0002; p = 0.0049 (c). d Pharmacodynamic biomarker levels tumors collected at end of study from animals treated in (b and c) (SMARCA4 protein: HCC515-vehicle (n = 8), HCC515-A947 (n = 10), HCC2302-vehicle (n = 10), HCC2302-A947 (n = 9); KRT80 transcript: HCC515-vehicle (n = 9), HCC515-A947 (n = 9), HCC2302-vehicle (n = 10), HCC2302-A947 (n = 8). Mice received a final dose of A947 24 h prior to tumor collection. SMARCA2 protein and KRT80 mRNA levels were quantified as in (a) and presented relative to levels in the respective vehicle-treated tumors. e Log-normalized counts per million reads (CPM) of SMARCA2 target genes determined by RNAseq and presented relative to levels in vehicle-treated tumors. A consensus set of SMARCA2-regulated genes (n = 412) were defined based upon exhibiting acute (24 h) and sustained (96 h) suppression following A947 treatment of both models in vitro, as further described in the methods section. Boxes span from quartile 1 (Q1; 25th perentile) to quartile 3 (Q3; 75th percentile), with medians (50th percentile) represented by center lines. Whiskers span ± 1.5 times the interquartile range (Q3–Q1) and outliers are represented as dots. f Tumor volume in mice harboring SMARCA4 wild-type Calu-6 xenografts following administration of A947 (40 mg/kg, i.v.) or vehicle control. Data are presented as mean ± s.e.m. (n = 10 mice/group). Potential significance was assessed by a two-sided, unpaired Student t test. n.s. nonsignificant. g Levels of SMARCA2, SMARCA4 and PBRM1 protein in Calu-6 tumors (n = 10 for each group) were quantified by ImageLab from Western blots and normalized to a loading protein control (β-actin). Calu-6 tumors were collected at the end of study from mice treated in (f) and mice received a final dose of A947 24 h prior to tumor collection. Source data are provided as a Source Data file.

Fig. 5. A947-mediated SMARCA2 degradation synergizes with MCL1 inhibition.

a Cell viability in 4 SMARCA4 mutant NSCLC cell lines measured across a small-molecule library of 723 compounds screened as a dose-response in the presence or absence of 100 nM A947 for 5 days. Data are plotted as the difference in the concentration required to inhibit growth by 50% (ΔIC₅₀) upon −/+ A947 treatment. Two separate MCL1 inhibitors are annotated in red. b Treatment of a representative SMARCA4-mutant cell line (NCI-H1793) with a 9 × 9 matrix titration of A947 with the MCL1 inhibitors, AMG-176 (upper plot) and S63845 (lower plot). Heatmaps depict activity in excess of the Bliss independence model to describe synergistic drug interactions (excess volume). c Live cell monitoring of apoptosis in SMARCA4-mutant NCI-H1944 (left graph) and NCI-H838 (right graph) cells grown in the presence of Caspase-3/7 Green Dye upon treatment with 100 nM A947 and/or 1 μM AMG-176. Data are presented apoptotic object count (mean ± s.d) in triplicate cultures. Significance was calculated by a two-, unpaired t-test; p = 0.0078 (NCI-H1944), p = 0.0002 (NCI-H838). Source data are provided as a Source Data file.