Targeting of BRM Sensitizes BRG1-Mutant Lung Cancer Cell Lines to Radiotherapy

Mutations of various subunits in the switch/sucrose nonfermenting (SWI/SNF) remodeling complex from cancer genome profiling were found in several cancers and are also common in lung adenocarcinomas (1, 2). The SWI/SNF complex, which consists of one of two mutually exclusive DNA-dependent ATPases BRG1/SMARCA4 or BRM/SMARCA2 together with core and accessory subunits, alters the chromatin structure to allow access of DNA binding proteins to double-stranded DNA that regulates essential cellular mechanisms such as transcription, replication, and repair (3). Hoffmann and colleagues and Oike and colleagues revealed BRM as a synthetic lethal target in BRG1-mutant cancers because of their selective growth inhibition after BRM depletion (4, 5).

Collisson and colleagues analyzed tumor and matched normal material from 230 previously untreated lung adenocarcinoma patients by whole-exome sequencing on tumor and germline DNA and observed that 6% of the tumors were mutated in SMARCA4. All major histologic types of lung adenocarcinoma were represented: 5% lepidic, 33% acinar, 9% papillary, 14% micropapillary, 25% solid, 4% invasive mucinous, 0.4% colloid, and 8% unclassifiable adenocarcinoma (2). Fukuoka and colleagues showed by IHC that between 30% and 50% of lung primary tumors have complete losses of BRG1 protein (6).

Recent studies have highlighted the importance of chromatin structure in the DNA damage response and showed how chromatin modifiers and remodelers, which relax the nucleosome structure, affect the DNA-damage response of the cells (7–9). It was proposed that nucleosomes pose a barrier to the processing of DNA ends, and this barrier must be relieved in order for end resection to occur (10).

The SWI/SNF complex as a chromatin remodeler is considered to be important for the processing of DNA DSBs. The yeast SWI/SNF complex, for example, facilitates the pairing of homologous DNA strands during homologous recombination repair (HRR; ref. 11). The mammalian SWI/SNF complex was also shown to regulate the DNA damage response (DDR). Ogiwara and colleagues have shown that BRM also regulates Ku70, a subunit of the DNA-PK repair complex mainly involved in end joining of DSBs and thus has an impact on nonhomologous end joining (NHEJ; ref. 12). BRG1, on the other hand, was shown to be recruited to DSBs by the retinoblastoma (RB) tumor suppressor to stimulate DNA end resection and HRR (13). Thus, targeting chromatin regulating proteins that are involved in DDR after IR may be a new strategy for individual radiosensitization of cancer cells.

The inactivation of the other mutually exclusive subunit BRM in BRG1-mutated tumor cells can lead to a synthetic lethal event as initially shown by using proliferation and colony-forming assays (4, 5).

To explore the effect of BRM knockdown on radiation response in non–small cell lung carcinoma (NSCLC), we investigated the clonogenic survival after single dose and tumor control in plaque-monolayer assays after fractionated irradiation as endpoints after IR treatment of NSCLC cells with and without BRG1 mutations. In addition, cell-cycle effects and DSB repair markers such as Rad51- and γH2AX/53BP1-foci were evaluated in order to determine by which pathway BRM silencing radiosensitizes NSCLC cells.

The human NSCLC cell lines A549, H1299, H661, HCC15, H460, H520, H596, and HCC827 were obtained from ATCC and maintained in RPMI-1640 (Invitrogen) supplemented with 10% FBS plus antibiotics, except A549 that was maintained in MEM (Invitrogen) supplemented with 15% FBS plus 1% nonessential amino acids and antibiotics. Adult fibroblasts (HDFA, PCS-201-012) and neonatal fibroblasts (HDFN, PCS-201-010) were obtained from ATCC. Fibroblasts were cultured with fibroblast basal medium (FBM, PCS-201-030 from ATCC) supplemented with low serum fibroblast growth kit (ATCC, PCS-201-041). NSCLC cell lines were usually propagated in medium supplemented with 5 μg/mL plasmocin (InvivoGen) as a Mycoplasma prophylaxis and were cultivated in an atmosphere consisting of 5% CO₂ and 95% air at 37°C. Experiments were done without plasmocin. All cell lines were usually re-thawed after 3 months in culture. Cells were irradiated using an X-ray machine RS320 (Xstrahl Ltd) at 300 kV, 10 mA, dose rate 0.9 Gy/minute.

Alterations in BRG1-coding regions were verified by Sanger sequencing in the cell lines A549, H1299, H661, and HCC15 as previously reported by other working groups (5, 14). Total RNA was extracted from four NSCLC cell lines using the RNeasy kit (Qiagen Inc.) as described earlier (14). The RNA was quantified by measuring absorbance at 260 nm/280 nm on a Nanodrop spectrophotometer (Thermo Fisher Scientific). To generate cDNA, 1 μg RNA was reversely transcribed using Superscript II reverse transcriptase kit (Invitrogen), according to the manufacturer's protocol. The specific PCR products were sequenced using Big Dye Terminator chemistry 3.1 (Applied Biosystems) with an ABI PRISM 3130 DNA Analyzer (PerkinElmer LifeSciences, Inc.). Identified BRG1 mutations in cDNA were further confirmed in the genomic DNA (Homo sapiens chromosome 19 reference assembly NC_000019.8). The A of the ATG translation initiation start site was considered as +1. The sequence of primers for PCR and DNA sequencing was designed with Primer3 software and are available upon request.

Generally, cells were cultivated in 9.6-cm² culture dishes at an initial density of 0.2–0.4 × 10⁶/well. Upon confluence of 70%–80% (20–24 hours of cell growth), transfection was carried out using 6 μL lipofectamine RNAiMax in serum-free Opti-MEM (both from Life Technologies). Transfection was applied with 50 nmol/mL specific silencer select siRNA (Life Technologies) for BRM (siRNA ID: s13133), nontargeting control siRNA #2 or without siRNA according to the supplier recommendations.

Cells were cultivated in 1.9 cm² culture dishes at an initial density of 0.1–0.2 × 10⁶ cells/well. After 20 to 24 hours of growth, transfection was carried out using 0.75–1.5 μL Lipofectamine 3000 and 1 μL P3000 reagent in serum-free Opti-MEM (both from Life Technologies). Transfection was applied with 0.5 μg control shRNA Plasmid-A (sc-108060, Santa Cruz) and BRM shRNA Plasmid (sc-29831, Santa Cruz) according to the supplier recommendations. After 72 hours, cells were treated with puromycin (2 μg/mL). Clones were picked and cultivated further with 2 μg/mL puromycin.

For reexpression of BRG1, cells were transfected with 0.25 μg pBRG1 (RC226420 SMARCA4, OriGene) using 1.5 μL lipofectamine 3000 and 1 μL p3000 reagent in serum-free Opti-MEM. After 72 hours, cells were treated with G418 (500 μg/mL). Clones were picked and cultivated further with 700 μg/mL G418.

Forty-eight hours after transfection, cells were plated at a cell density of 2 × 10⁵ cells/T25 flask. Cell number was determined after 24, 48, and 72 hours by automated cell counting system Luna (Logos Biosystems). For the measurement of clonogenic survival, cells were harvested 48 hours after transfection and plated in triplicate in 9.6-cm² culture dishes. After 4 to 6 hours in culture, cells were irradiated and incubated at 37°C in 5% CO₂ for 10–14 days. Cells were fixed and stained [with 96% ethanol, 15% (w/v) Giemsa] and destained with distilled water. Colonies consisting of at least 50 cells were counted. The surviving fractions after the respective radiation doses are presented as a fraction of the growth of untreated colonies.

Cells were harvested and lysed in RIPA buffer (Thermo) supplemented with a 1× protease inhibitor cocktail (Roche) for 15 minutes at 4°C. Uniform amounts of protein extracts were mixed with 4× LDS sample buffer (Invitrogen) and heated to 95°C for 5 minutes prior to gel loading. Separation of proteins was carried out using NuPAGE 4%–12% bis–tris or 7% tris–acetate gels (Invitrogen). After electrophoresis, proteins were transferred to Invitrolon PVDF blotting membranes (Invitrogen). Then, the membrane was blocked in 5% defatted powdered milk/Tris–buffered saline with 0.5% Tween 20 for 30 minutes at room temperature and incubated with the primary antibody overnight at 4°C. After washing, the membrane was incubated with the secondary antibody conjugated with horseradish peroxidase or with Alexa-488. Protein visualization was performed using SuperSignal West Femto Stable Peroxide Buffer and Luminol/Enhancer Solution (Thermo). Signal detection took place using ChemiDOC MP Imaging System (Bio-Rad). The following antibodies were used: rabbit mAb anti-BRM 1:1,000 (D9E8B, Cell Signaling); rabbit Ab anti-BRG1 1:1,000 (A52, Cell Signaling), anti-rabbit IgG HRP-linked antibody 1:2,000 (7074P2 Cell Signaling), as loading control rabbit pAb to GAPDH (Abcam, ab9485) or mouse mAb to GAPDH (abcam, ab8245) were used diluted 1:2,000, as secondary antibody Alexa Fluor 488 goat anti-mouse IgG 1:2,000 (Life Technologies, A11017) was used.

After siRNA transfection for 48 hours, cells were trypsinized and reseeded in chamber slides and were irradiated 4 hours later. Cells were washed and fixed at the indicated times after IR in 4.5% formaldehyde, washed with PBS followed by a treatment in permeabilizing buffer (100 mmol/LTris-HCl, pH 7.4; 50 mmol/L EDTA, 0.5% Triton X-100) for 15 minutes at RT and washed in PBS. After incubation in blocking buffer (3% BSA, 0.1% Tween 20, 4× SSC), primary antibodies rabbit anti-Rad51 1:500 (Calbiochem, PC130-100), rabbit anti-53BP1 1:500 (ab21083), mouse anti γ-H2AX 1:500 (ab22551) were applied to the cells and incubated overnight. The secondary antibodies goat anti-rabbit Cy3-conjugated antibody 1:500 (Jackson ImmunoResearch, 800-367-52) and Alexa Fluor 488 goat anti-mouse 1:500 (Life Technologies, A11017) were applied together with 4′,6-diamidino-2-phenylindole DAPI nuclear counterstain 1 μg/mL for 1.5 hours at room temperature. Slides were examined on a fluorescence microscope Imager Z1 (Zeiss).

The plaque-monolayer assay was previously described (15). Briefly, 1,500 cells were seeded as a small plaque in 5 μL medium as a 3-mm spot into each well of a 2-cm² 24-well culture plates with about 80% of the cells having cell–cell contacts. Cells were irradiated 24 hours after seeding. The number of plaque monolayers reaching the survival criteria was determined weekly and monitored for up to 41–45 days after irradiation. A plaque monolayer was designated as surviving if cells reached >50% confluency or with >10-fold doubling of the initial cell number. The monolayer control rate was calculated as the ratio of nonproliferating cultures and the total amount of seeded monolayers/treatment. The proportions of plaque-monolayer control in the different treatment groups were analyzed by logistic regression.

Animal experiments were conducted according to the German animal welfare regulations and approved by the local authorities (Az. 84-02.04.2014.A481; Az. 81-02.04.2018.A158; Az. 84-02.04.2013.A453). Immunodeficient NMRI (nu/nu) nude mice were purchased from the Central Animal Facility (ZTL) of the University Hospital Essen (age 8–16 weeks). Animals were housed in an individually ventilated cage rack system (Techniplast) and fed ad libitum. Cells derived from the human NSCLC lines A549 and H460 were trypsinized and collected after transfection with siRNA targeting BRM or nontargeting control siRNA #2 (Life Technologies). Cells were centrifuged, resuspended, and diluted in media without fetal bovine serum or antibiotics to a concentration of 2 × 10⁶ cells per 200 μL. Cells were subcutaneously injected in a volume of 200 μL into the flank of NMRI mice. At 14 days later, when the tumors reached a volume of 180–300 mm³, they were explanted and mechanically minced with scalpels and enzymatically dissociated with (5U/mL) dispase (STEMCELL Technologies) and (5 μg/mL) collagenase type II (Merck Millipore), for 1 hour, filtered through a 100-μm filter and seeded onto glass chamber slides in the respective growth medium. Cells were irradiated at 4 hours later with 10 Gy and fixed at 24 hours thereafter in 4.5% formaldehyde and processed for immunofluorescence detection of Rad51-foci in cyclin B1-positive cells. Foci were counted in 400 cells per treatment group.

Colony data from each cell line were analyzed using a general linear model with type of siRNA transfection, i.e., siBRM, nontargeting siRNA, lipofectamine without siRNA (transfection), individual repeated experiment (experiment, at least three repeats), and radiation dose (dose) as classification variables (ProcedureMIXED, SAS/STAT 14.1, SAS Institute Inc.). The dependent observations were the logarithms of the ratio of counted colonies versus seeded cells from each experiment with at least three cell culture plates. The ratios of counted colonies to seeded cells were normalized for each experiment to the plating efficiency at 0 Gy using the same transfection condition. The independent variables of the model were the radiation dose, the transfection condition, the transfection condition crossed by the radiation dose level, and the individual experiment crossed by the radiation dose level. The experiment × dose effect was declared as a random effect. The significance of the effect was assessed by an ANOVA F-test. Repeated-measures multivariate analysis of the colony survival data given in Table 2 repeated at the different dose levels was performed with procedure GLM, SAS. Assessing a hypothesis in multiple cell lines, a significant effect in a cell line was claimed only if the associated P value was smaller than the significance levels after Bonferroni-type correction (16).

Plaque-monolayer control curves were analyzed as described previously (15). Logistic regression was performed using the Procedure Logistic, SAS. The radiation dose effect was introduced as a continuous variable, the transfection group, lipofectamine, noncoding siRNA, and BRM siRNA was introduced as a classification variable as well as an experiment-wise deviation of the radiation responsiveness.

To examine the impact of BRM depletion, we compared the effect of siRNA-mediated BRM knockdown on BRG1-deficient and BRG1-proficient cells with respect to their short-term proliferation potential. As shown in Table 1, BRG1 mutation screening of NSCLC cell lines confirmed the mutation status of the cells as previously reported (14). These mutations lead to loss of the full-length BRG1 protein (Fig. 1A, bottom). All cell lines tested express BRM, albeit with different expression level. Relative expression level in the different cell lines was compared with the BRG1-deficient A549 line with the highest expression of BRM. HCC827 cells show a low expression level less than 1% in comparison with A549 (Fig. 1A, top). The highest BRG1 expression level was detected in H520 cells (Fig. 1A, bottom).

Specific siRNA targeting BRM reduced its expression to about 24% ± 6% in H1299, 30% ± 6% in A549, 36% ± 6% in H661, 46% ± 8% in HCC15, 6% ± 2% in H460, 30% ± 7% in H520, 28% ± 7% in H596 and 3% ± 1% in HCC827 of the protein level compared with the cells treated with control siRNA (Fig. 1B). Depletion of BRM protein significantly inhibited proliferation of BRG1-deficient but not of BRG1-proficient cells (Fig. 1C). In H661 cells, proliferation was also inhibited by control siRNA. Therefore, the impact of siBRM on proliferation of H661 cells cannot be determined in this assay.

To examine the impact of BRM depletion on long-term survival, colony formation assays were performed. BRM knockdown in cell lines with BRG1 mutations led to a significant reduction of plating efficiency in comparison with control transfected cells, except H661, where the treatment with control siRNA also led to a reduction of the plating efficiency (Table 2). However, shRNA control-transfected cells showed a smaller effect on plating efficiency in H661 than control siRNA (Table 2). In comparison, BRM knockdown did not affect the plating efficiency of NSCLC cells with wild-type BRG1 nor human fibroblasts (Table 2). Taken together, the results from short-term proliferation and clonogenic survival assay showed significant inhibitions of proliferation and clonogenicity after BRM knockdown specifically in cell lines with mutated, i.e., nonfunctional BRG1 but not in cell lines with wild-type BRG1.

Colony formation assays were performed in BRG1-deficient and BRG1-proficient cell lines to evaluate the impact of BRM knockdown on radiosensitivity of the NSCLC cell lines. In addition, adult (HDFA) and neonatal (HDFN) skin fibroblast cells were used as a control for nonneoplastic cells. Nontarget siRNA transfection did not change the surviving fractions in each of the cell lines in comparison with exposure with the transfection reagent (lipofectamine) alone (P > 0.2, F-test, Bonferroni corrected for multiple testing) and therefore data from both exposures were combined as controls (Fig. 2). Knockdown of BRM increased the radiation sensitivity in all BRG1-deficient NSCLC cells (Fig. 2A) but not in BRG1-proficient NSCLC (Fig. 2B) or fibroblast cells (Fig. 2C). Knockdown of BRM with siRNA reduced the surviving fraction of irradiated cells significantly in comparison with the control cells over radiation dose levels between 2 and 6 Gy for each BRG1-deficient cell line as compared with the controls (P < 0.0001, F-test). The average radiation sensitizing effect of BRM knockdown on all BRG1-mutated tumor cell lines at radiation doses between 2 and 6 Gy can be described by a survival reducing factor of 0.63 (95% CI, 0.57–0.69). The average radiation-sensitizing effect of BRM knockdown over all BRG1-proficient tumor cell lines at radiation doses between 2 and 6 Gy can be described by a survival reducing factor of 1.03 (95% CI, 0.90–1.18) with some heterogeneity among BRG1-proficient cell lines (Table 3). The survival modifying factors due to BRM knockdown were significantly smaller in BRG1-deficient than in BRG1-proficient NSCLC cells (P = 0.008, F-test; Table 3). Because control siRNA oligonucleotides had some effects on plating efficiency and proliferation in H661, survival curves were generated also after BRM knockdown with shRNA in comparison with sh-control-transfected H661 cells. The survival reducing factor of BRM knockdown with shRNA was 0.64 (95% CI, 0.51–0.80) in close agreement to the results obtained when using siRNA (Table 3). To determine if IR sensitization is dependent upon BRM knockdown (Supplementary Fig. S1B) in a BRG1-deficient background, clonogenic survival has been measured in cells with restored expression of BRG1 in H1299. The data from 4 independent H1299 clones (Supplementary Fig. S1A) showed that the radiosensitization effect of siBRM in H1299 is dependent on BRG1 (Supplementary Fig. S1C). BRG1 reexpressing H1299 clones (Supplementary Fig. S1A) did not show a significant radiation-sensitizing effect of BRM knockdown at radiation doses between 2 and 4 Gy. The average survival-reducing factor at radiation doses between 2 and 4 Gy was 0.91 (95% CI, 0.83–1.01; P = 0.064).

Comparing the surviving fractions of BRG1-mutated and -proficient cell lines using repeated measurements at 2, 4, and 6 Gy (Table 3) showed that BRG1-mutated cell lines had higher surviving fractions than BRG1-proficient cell lines in this panel of cell lines (P = 0.009, F-test), indicating an effect of the functional BRG1 status on radiation sensitivity in human wild-type NSCLC lines.

The effect of fractionated irradiation schedules up to total doses that permanently control the growth of plaques consisting of 1,500 tumor cells was investigated in the two most radioresistant BRG1-mutated and two BRG1-proficient cell lines in the clonogenic assay. Cells were irradiated with one fraction (4 Gy) per day, 7 fractions a week (with minimal interval of 24 hours) to the respective total irradiation doses. The maximum overall treatment time to the highest total dose of 48 Gy was 12 days, which allows cells to repopulate during this time. Importantly, BRM downregulation was stable at least for 8 days as measured in H1299 with and without irradiation at the respective TCD with 15 Gy (Supplementary Fig. S2). In addition, repeated application of 4 Gy can augment a sensitizing effect of BRM knockdown. Direct effects of transfection on the plating efficiency, which is independent from the radiation response, can also influence the plaque-monolayer control. Figure 3 shows the influence of BRM depletion on the radiation dose–response relation for plaque-monolayer control for the BRG1-mutated cell lines A549, H1299, and wild-type BRG1 cell lines H460 and H520 in comparison with the transfection reagent and nontargeting siRNA-transfected controls.

Transfection with nontargeting or BRM siRNA had no effect on the location of the plaque-monolayer control curve for the BRG1-proficient cell lines H460 (P = 0.59, F-test) and H520 (P = 0.49, F-test). The total radiation doses that control 50% of the plaque monolayers (TCD₅₀ values) were 11.1 ± 0.2 Gy and 16.7 ± 0.4 Gy for the noncoding siRNA controls that were shifted by +0.3 ± 0.2Gy and +0.4 ± 0.5 Gy due to BRM siRNA transfection in the H520 and H460 lines, respectively. In comparison, the plaque monolayers of A549 and H1299 with deficient BRG1 were more resistant compared with H460 and H520 with wild-type BRG1. The respective TCD50 values for the noncoding siRNA transfection group were 38.4 ± 0.5 Gy and 27.9 ± 0.2 Gy, for A549 and H1299, respectively. Transfection with BRM siRNA shifts the TCD₅₀ value for A549 by −3.8±0.5 Gy and with lipofectamine by + 2.9 ± 0.7 Gy, respectively. All three transfection groups (i.e., lipofectamin, siControl, and siBRM) differ in the pairwise comparisons (P < 0.0001, F-test). We observed the same holds for H1299 with TCD50 changes of −3.8 ± 0.3 Gy and +2.7 ± 0.4 Gy by BRM siRNA transfection or lipofectamine alone exposure. The most resistant cell line in the minimonolayer model, A549 was also studied after single dose irradiation. The TCD₅₀ for single dose irradiation in the noncoding siRNA transfection group was 14.5 ± 0.2 Gy. TCD₅₀ was decreased in the BRM siRNA transfection group by −1.9 ± 0.5 Gy and increased in the lipofectamine group by +1.3 ± 0.4 Gy (P < 0.0001, χ² test).

Staining of Rad51-foci as a marker for HRR was done at 4 and 24 hours after irradiation following BRM silencing. In BRG1-mutated cell lines the initial number of Rad51-foci at 4 hours after IR was not changed but the residual number of Rad51-foci at 24 hours after IR significantly increased after BRM knockdown in the whole population of cells (Fig. 4B) as well as in cells that where counterstained with Cyclin B1 as a marker for G₂ phase cells (Fig. 4C). To exclude, that the increase of Rad51-foci at 24 hours after irradiation is due to an increase of S–G₂ phase cells, cell-cycle phase distribution at the indicated time point (24 hours) after irradiation was studied. Overall, cell-cycle profiles did not reveal significant changes after BRM knockdown in an exemplarily tested BRG1-deficient cell line A549 24 hours after irradiation in comparison with control (Supplementary Fig. S4). The depletion of BRM in cell lines with wild-type BRG1 revealed no differences in Rad51-foci number at 4 hours (initial level) or at 24 hours (residual level) after irradiation compared with the siRNA control-treated cells (Fig. 4A). Depletion of BRM in cell lines with restored BRG1 expression also revealed no differences in Rad51-foci number at 4 hours after 4 Gy (P = 0.13, t test) or at 24 hours after 4 Gy or 10 Gy irradiation (P > 0.05, t test) compared with the siRNA control-treated cells (Supplementary Fig. S1D).

In addition, we used γ-H2AX and 53BP1-foci staining at 1, 4, and 24 hours after irradiation as a measure for DNA damage associated with NHEJ (Supplementary Fig. S3A–S3B). Neither BRG1 wild-type (Supplementary Fig. S3A) nor BRG1-mutated cell lines (Supplementary Fig. S3B) showed an effect of BRM knockdown on γ-H2AX or 53BP1-foci formation.

Residual numbers of Rad51-foci at 24 hours after irradiation were also analyzed in ex vivo explant cultures of xenograft tumors subcutaneously grown on NMRI nu/nu mice from cells derived from BRG1-mutant (A549) and wild-type (H460) cells (Supplementary Fig. S5A) after transfection with siRNA targeting BRM or nontargeting control (Supplementary Fig. S5B). Depletion of BRM significantly increased the number of background corrected residual Rad51-foci of 3.91 ± 0.88 from a mean number of 4.14 ± 0.63 to 8.04 ± 0.63 at 24 hours after irradiation with 10 Gy in A549 (P < 0.0001, F-test). Depletion of BRM had no significant effect on residual number Rad51-foci in H460, with a decrease in the number of background corrected residual foci of −1.10 ± 0.86 foci (P = 0.27, F-test) from a mean number of 8.75 ± 0.61 to 7.65 ± 0.61 Rad51-foci at 10 Gy with control and BRM siRNA, respectively (Supplementary Fig. S5A). These data showed the potential transferability of the in vitro results to at least ex vivo tumor models.

Identification of cancer-specific vulnerabilities created by mutations in one of the subunits of the SWI/SNF complex resulting in a synthetic lethal relationship can be of substantial relevance to cancer therapy. The first finding of a synthetic lethal interaction that has been translated into the clinic was the sensitivity of BRCA1/BRCA2-deficient tumors toward PARP inhibition (17). In the present study, the role of SMARCA4/BRG1 mutations combined with SMARCA2/BRM knockdown was evaluated for the radiation sensitivity of human NSCLC cell lines.

Our results show that downregulation of BRM significantly decreased cell proliferation and plating efficiency of BRG1-mutated NSCLC cells, which was also previously described by Oike and colleagues (5). To address the potential significance of BRM-inhibitory therapies in combination with radiotherapy, we examined the radiation sensitivity measured by clonogenic survival of BRG1-deficient and -proficient NSCLC cells. This is the first report indicating increased radiation sensitivity specifically in BRG1-mutant but not BRG1 wild-type cells and fibroblasts as a surrogate for nontransformed cells after depletion of BRM.

The results of the plaque-monolayer assay, which is a tumor population survival assay, reveal that BRM depletion specifically increases the radioresponse of BRG1-deficient NSCLC cells after fractionated as well as single dose irradiation in comparison with the transfection reagent alone and nontarget siRNA control. Minimonolayers of BRG1-mutant tumor cell lines A549 and H1299 become rather radioresistent after fractionated daily irradiation of 4 Gy with TCD₅₀ values of >25 Gy for the siRNA controls. Additional resistance-promoting effects such as repopulation and sublethal damage repair, which are not present during single dose irradiation, are triggered after fractionated radiation (15, 18). The surviving fraction at 4 Gy has to be raised to a higher power during multifraction irradiation to control the minimonolayers of the more resistant BRG1-mutant cell line. This sensitive assay did not only show the effects of BRM siRNA transfection on radiation response, but also smaller effects of control siRNA transfection became detectable: the latter is not discovered in the cell population growth or the clonogenic assay. Nonspecific effects of transfection with noncoding siRNA can rely on the activation of the type I interferon (IFN), JAK/STAT signaling pathway and of IFN-responsive genes (19, 20). Specifically, the RNA-dependent protein kinase (PKR) has previously been shown to mediate the type I IFN-dependent effects of siRNAs, which affect biological processes including cell growth, differentiation, and apoptosis. Double-stranded RNA was shown to be the most well-characterized activator of PKR (21). It was shown that various pathways including immune response, IFN-related genes, DNA damage, cell-cycle arrest, TGFβ, survival, and apoptotic signal transduction can be modified by fractionated irradiation (22, 23) and thereby potentiates small nontarget effects of siRNA.

To further analyze the underlying mechanism of the sensitizing effect after BRM depletion, we evaluated the role of BRM and BRG1 on Rad51-foci formation, a marker of HRR, after irradiation. A decreased resolution of Rad51-foci 24 hours after BRM knockdown was found only in BRG1-mutated but not in wild-type cell lines. Activity of the HRR is restricted only to S- and G₂–M phases of the cell cycle (24). In order to exclude the possibility that an increase in S- and G₂-phase cells after BRM knockdown may be responsible for the increased number of Rad51-foci, cell-cycle analysis at 24 hours after IR were performed. There was no effect of siBRM on cell-cycle progression within 24 hours after irradiation. In addition, Rad51-foci analysis in S–G₂ phases by costaining with Cyclin B1 shows the same results, with increased Rad51-foci at 24 hours after irradiation in BRG1-deficient cell lines transfected with siBRM. Of importance is the finding that these results of a selective BRM dependency of the radiation response of BRG1-mutant cancers could be confirmed in ex vivo explant cultures from xenograft tumors on nu/nu mice. This has the potential of a functional predictive assay applicable to fresh biopsies.

Evidence exists that SWI/SNF is involved in HRR. Depletion of BRG1 was found to result in defective Rad51 filament assembly (25).The displacement of yeast heterochromatin factors by the SWI/SNF remodeler complex was also shown to promote Rad51-mediated joint-molecule formation and RAD54-dependent strand invasion, priming the DSB for repair by HRR (26). Rad54 is a protein of the SWI2/SNF2 complex with ATPase activity, which requires the presence of dsDNA and interacts with Rad51 to stabilize Rad51 filaments and is involved in strand invasion (27). In addition, it was shown that the SWI/SNF complex can orchestrate the recruitment of Mre11–Rad50–Xrs2 that is specifically dedicated to HRR (28).

It was also proposed that the SWI/SNF complex can have a role in nonhomologous recombination repair (NHEJ). SWI/SNF complexes were shown to facilitate DSB repair by promoting H2AX phosphorylation (29, 30). BRM might be required for recruitment of the KU70/80 subunits of the DNA-PK repair complex to damaged DNA and thus for efficient NHEJ (12). However, the data from the present study did not support an involvement of BRM in radiation-induced DNA double-strand signaling via γH2AX or 53BP1 as a measurement for early repair of DSB, because BRM knockdown has no significant effect on initial and residual γH2AX as well as of 53BP1-foci formation at 1, 4, and 24 hours after irradiation.

The exploratory analysis of four cell lines with and the four cell lines without BRG1 mutations shows that BRG1-mutated cell lines are significantly more resistant to ionizing radiation than cell lines without BRG1 mutation. Thus, BRG1 expression could be a potential biomarker for the radiation sensitivity of NSCLC, a hypothesis that should be further analyzed in clinical series comparing local tumor control after definitive radiochemotherapy and BRG1 mutations in the respective tumors. Inhibition of BRM-containing residual complexes could be a promising approach for the treatment of BRG1-mutant cancers justified by results from previous studies: 30% to 50% of lung primary tumors have complete losses of BRG1 protein (31). Besides, mutations in the BRG1 gene are frequently found in human lung adenocarcinomas (2), and decreased SMARCA4/BRG1 expression is associated with worse prognosis after chemotherapy in patients harboring lung adenocarcinomas (32).

BRM is a potentially druggable target, which harbors at least two targetable domains, an enzymatic ATPase domain and a bromodomain. Small molecules targeting BRM are now under development and available, e.g., ADAADi for the ATPase domain of BRM (33–36). In addition, bromodomains have recently emerged as particularly good targets for anticancer therapy (37) and some small molecules are under development, such as PFI-3 (37, 38). BRM could also be inactivated indirectly, as for example via C-terminal acetylation by HDAC2 inhibitors (39). In addition, functional inhibitors of BAF complex containing BRM in the case of BRG1 mutation were identified by high-throughput screening of small-molecule libraries (40).

Although the development of BRM inhibitors is at an early stage, the vulnerability of tumors with BRG1 mutations was found to be increased with respect to cell death induction or growth inhibition toward inhibitors of other protein classes apart from BRM, such as Aurora kinase A or EZH2 (41, 42). Small-molecule inhibitors are available for the latter targets that offer the opportunity to deliver a tumor-selective therapy for BRG1-mutated tumors in the clinic. However, radiation-sensitizing effects of these inhibitors in the background of BRG1 deficiency were not studied up to now. In conclusion, the present study shows that SMARCA4/BRG1 mutations are associated with decreased radiation responsiveness in human NSCLC cell lines, which can be increased upon SMARCA2/BRM depletion and thus identifies BRM as an interesting therapeutic target in SMARCA4/BRG1-mutant cancers.

No potential conflicts of interest were disclosed.

Conception and design: E. Zernickel, A. Sak, M. Stuschke

Development of methodology: A. Sak, A. Riaz, M. Groneberg, M. Stuschke

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Sak, D. Klein, M. Groneberg, M. Stuschke

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Zernickel, A. Sak, M. Stuschke

Writing, review, and/or revision of the manuscript: E. Zernickel, A. Sak, A. Riaz, D. Klein, M. Stuschke

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Zernickel

Other (performed experiments): A. Riaz

We thank George Iliakis for valuable discussions, reagents, and technical advice. This work is funded by the Deutsche Forschungsgemeinschaft (DFG) as part of the Graduate School (GRK1739/2).