Autophagy-Dependent Sensitization of Triple-Negative Breast Cancer Models to Topoisomerase II Poisons by Inhibition of the Nucleosome Remodeling Factor

Because epigenetic regulators play a prominent role in cancer-specific biology, they have been targeted to develop novel cancer therapies (1). Several laboratories, including ours, have discovered that inhibiting the chromatin remodeling complex nucleosome remodeling factor (NURF) suppresses cancer growth using in vitro and in vivo models (2–7). NURF is a prototypical chromatin remodeling complex, catalyzing the ATP-dependent sliding of nucleosome positions in chromatin with profound effects on gene expression (8).

Doxorubicin and etoposide are widely effective chemotherapeutic agents that act by targeting topoisomerase II (TOP2). Mammalian TOP2 resolves DNA topologic issues such as supercoils and catenanes by cleaving the DNA duplex, generating the catalytic intermediate TOP2 cleavage complex (TOP2cc; ref. 9). Doxorubicin and etoposide induce cellular damage primarily through the stabilization of TOP2cc (hence they are termed TOP2 poisons), which interfere with DNA metabolism and, if left unrepaired, lead to cell death (10). To survive, cancer cells in part activate autophagy to consume and recycle damaged macromolecules and generate energy. Targeting autophagy pharmacologically to sensitize tumor cells to doxorubicin has been demonstrated with a variety of cell line models (11).

Autophagy is a well-characterized cellular process induced in cancer cells after exposure to cytotoxic therapies (12). Autophagy is regulated by a well-characterized pathway requiring 30 or more autophagy-related proteins (ATG), which leads to the sequestering of select cargos into the autophagosome. The autophagosome fuses with a lysosome where the low pH and degradative enzymes digest the cytoplasmic components to micromolecules, thereby providing nutrients to promote survival (12). Autophagy occurs in many forms including macroautophagy, microautophagy, chaperone-mediated autophagy, mitophagy, and lipophagy. These processes differ in the type of cargo targeted for degradation by fusion with the lysosome. In the case of mitophagy, defective mitochondria are targeted for degradation, which improves cell homeostasis in times of metabolic stress (13).

In this study, we examined whether NURF regulates sensitivity to commonly used anticancer agents by inhibiting its essential BPTF subunit. From our studies, we discovered that BPTF suppresses the therapeutic effects of a TOP2 poison's ability to enhance TOP2-induced DNA damage and it appears to require functional autophagy.

Wild-type (WT) and BPTF knockdown (KD; described in ref. 2) 4T1, 66cl4 4T07 and 67NR (Gift from Fred Miller, Wayne State University, Detroit, MI), MDA-MB-231 (ATCC, catalog no. HTB-26) and HEK 293T (ATCC, catalog no. CRL-3216) were cultured in complete media (CM; DMEM, 10% FBS, 2 mmol/L glutamine, and penicillin/streptomycin). MCF7 cells (ATCC, catalog no. HTB-22) were grown in CM with 0.01 mg/mL human insulin. Embryonic stem (ES) cells were maintained as described previously (14). Cell lines were used within 4 passages and when not obtained from ATCC were authenticated by short tandem repeat profiling.

shRNAs were introduced into MDA-MB-231 cells using the lentiviral pLVTHM system. pLVTHM plasmids Ctrl-sh1, Bptf-sh1, Bptf-sh2 are available at Addgene as stock numbers 83046, 83277, 83278, respectively. Transduced cells were selected with 0.5 μg/mL puromycin after 48 hours and subsequently grown in as described above.

Atg5 shRNA was introduced into the WT and BPTF KD 4T1 cells using the MISSION pLKO.1 lentivirus system (Millipore-Sigma). Viruses were concentrated by ultracentrifugation as described previously and used immediately to transduce WT and BPTF KD 4T1 cells lines without selection. Atg5 KD was confirmed 72 hours posttransduction by Western blotting.

Mycoplasma contamination was tested and confirmed to be negative every year using Universal Mycoplasma Detection Kit (ATCC, catalog no. 30–1012K).

Cells were plated into 6-well plates and the following morning cells were treated with chemotherapy or radiation. Chemotherapies were diluted in culture media Cells were treated for 24 hours after which they were washed and cultured in fresh media. Colonies were stained with crystal violet and counted. The fraction viable after treatment was calculated by the ratio of no. colonies treated/no. of colonies untreated.

To measure doxorubicin accumulation cells were treated with 1 μg/mL doxorubicin for 1 hour. Cells were washed, dispersed from culture, washed twice in FACS buffer (1% FBS, Hank's balanced salt solution; HBSS), resuspended, and analyzed. To measure rhodamine 6G accumulation, cells were treated with 1 μmol/L rhodamine 6G for 3 hours and processed as described above. Efflux was measured by washing cells twice with fresh media after and culturing for 24 hours. Media was changed once during the 24-hours Hoechst 33258 was performed similarly including 1-hour accumulation of 15 μmol/L and 24-hour efflux.

Cells were treated with drugs for 24 hours. After drug removal cells were collected, washed with PBS and centrifuged. Pellets were suspended in 90% ice-cold methanol fixed and permeabilized for 30 minutes at room temperature. Then, cells were washed in PBS again, suspended in 2% BSA, and blocked for 30 minutes on shaker. γH2AX-conjugated antibody (BD Pharmigen, catalog no. 560445) was added 1:100 and cells were stained for 1.5 hours on shaker protected from light. Cell were washed and analyzed.

Apoptosis was monitored by Annexin V-FITC/PI staining (BD Biosciences) according to the manufacturer's instructions. Samples were analyzed by BD FACSCanto II and BD FACSDiva software at Virginia Commonwealth University Flow Cytometry Core Facility as described previously.

After exposure to chemotherapy or radiation, and recovery, cells were stained with 1 μg/mL acridine orange (Invitrogen, catalog no. A3568) at 37°C for 25 minutes and then washed with PBS. Cells were observed under an inverted fluorescence microscope and images taken at 20× magnification. For quantification of autophagy, cells were collected with trypsin and washed with PBS and analyzed by flow cytometry.

Cells were seeded in 6-well plates on coverslips. After exposure to chemotherapy or radiation, cells were stained with 100 nmol/L LysoTracker Red DND-99 (Invitrogen, L7528) at 37°C for 30 minutes and then 100 nmol/L MitoTracker Green FM (Invitrogen, M7514) for an additional 30 minutes. Cells were observed under a fluorescence microscope and images taken at 100× magnification. Overlapping LysoTracker-positive and MitoTracker-positive organelles were determined by image overlay.

To quantify β-galactosidase (β-Gal)-positive senescent cells, after exposure to doxorubicin, cells were treated with bafilomycin (100 nmol/L) for 1 hour to achieve lysosomal alkalinization and stained with C₁₂FDG (Invitrogen, catalog no. D2893; 10 μmol/L) fluorescent β-Gal substrate for 2 hours at 37°C. After incubation, cells were collected, washed, suspended in PBS and analyzed by flow cytometry. All experimental procedures were performed with cells protected from light.

BALB/cJ (Jackson Labs, catalog no. 000651) or NSG (Jackson Labs, catalog no. 005557) mice 6–8 weeks of age weighting approximately 20 g were housed under aseptic barrier conditions as approved by Virginia Commonwealth University IACUC animal protocol AD10000017. 5 × 10⁴ 4T1 cells were surgically transplanted into the fourth mammary fat pad of BALB/c mice. 2 × 10⁶ MDA-MB-231 cells were introduced into the flank. Tumors were measured by caliper measurements 3 days/week (BALB/c) or 1 a week (NSG) starting from days 7 posttransplantation. Doxorubicin (or PBS control) at 5 mg/kg was given to BALB/c mice once a week intraperitoneally and at 3 mg/kg once a week to NSG mice intravenously.

Alamar blue assay (Thermo Scientific, catalog no. DAL1025) was done with the MDA-MB-231-spheroids seeded on 96-well spheroids plate (Corning). Spheroids were treated with 1 μmol/L doxorubicin for 4 days and 10% Alamar blue in media was used to develop the purple color for 24 hours and the fluorescence was measured on a microplate reader. A similar assay in parallel was done for the untreated cells. The percentage (%) difference of the live cells between control and treated was calculated.

TOP2ccs were isolated and detected using in vivo complex of enzyme (ICE) assay as described previously (15). Briefly, cells were lysed to reduce the viscosity of DNA and layered onto CsCl solution (150% w/v), followed by centrifugation. The resulting pellet containing was dissolved in TE buffer. Two micrograms of DNA is applied per slot on the blotting apparatus and was subsequently subjected to immunoblotting with TOP2α (Abcam, catalog no. ab52934) and TOP2β (Novus Biologicals, catalog no. NB-100–40842). DNA loading was measured using an anti-DNA antibody (Abcam, ab27156). Quantitative slot blotting was performed using Pierce Super Signal ELC Femto detection reagent (Thermo Fisher, catalog no. 37075) with the Bio-Rad ChemiDoc System. Top2cc and DNA signals were quantified by densitometric analysis using ImageJ.

After 20-minute doxorubicin (100 μmol/L) treatment, 4T1 cells were subjected to subcellular fractionation to isolate chromatin using Subcellular Protein Fractionation Kit (Thermo Fisher, catalog no. 78840) following manufacturer's instructions for Western blotting to assess the chromatin fraction of TOP2α and β.

Total proteins from cell cultures was extracted by TRI Reagent (Millipore-Sigma, catalog no. T9424) using manufacturer-suggested protocol. Protein pellets were dissolved in 8 mol/L urea and 1% SDS at 65°C then quantified using the DC protein assay (Bio-Rad, catalog no. 5000112) and a BSA standard. Proteins were resolved by SDS-PAGE, transferred to polyvinylidene difluoride as described previously (16). The following primary antibodies were used: anti-BPTF (Millipore-Sigma, ABE24), anti-SNF2 L (Santa Cruz Biotechnology, sc-79088), pRbAp46/48 (Santa Cruz Biotechnology, sc-33170), TOP2α (Abcam, ab52934), TOP2β (Novus Biologicals, NB-100–40842), LC3B (Novus Biologicals, NB100–2220) ATG5 (Cell Signaling Technology, 12994S), anti-Cyclophilin B (Thermo Fisher Scientific, PA1–027A), anti-GAPDH (Cell Signaling Technology, 2118), then anti-mouse or rabbit HRP secondary (Cell Signaling Technology, 7074S). Qualitative Western blotting was done using film. Quantitative Western blotting was performed using Pierce Super Signal ELC Femto detection reagent (Thermo Fisher Scientific, 37075) with the LI-COR Odyssey Fc system. Loading was determined by Cyclophilin B or GAPDH.

Excel was used to calculate statistical differences between groups using a paired two-tailed Student t test. When required, t test P values were adjusted using the FDR method in R software for all figures involving multiple comparisons'.

To create a NURF-inhibited mouse triple-negative breast cancer (TNBC) 4T1 cell line, BPTF was targeted for KD because it is unique and essential to the function of the NURF complex (Fig. 1A; ref. 17). Western blotting confirmed that BPTF shRNAs depleted BPTF, but not the SNF2 and pRbAp48 subunits, in 4T1 cells (Fig. 1B). BPTF KD 4T1 cells were assayed for sensitivity to 7 chemotherapeutic drugs and ionizing radiation through clonogenic survival assays. These studies indicated that BPTF KD sensitized 4T1 cells to the TOP2 poisons, doxorubicin and etoposide, the microtubule poison paclitaxel, and the histone deacetylase inhibitor, (HDACi) panobinostat, but did not sensitize the cells to the DNA-damaging agents, ionizing radiation or cisplatin, the DNA methyltransferase inhibitor (DNMTi) azacytidine, or the nucleoside analogue 5-fluorouracil (Fig. 1C).

To examine the selectivity of BPTF as a suppressor of chemotherapeutic drug sensitivity for cancer cells, we also assayed BPTF KO embryonic stem cells (ESC) as a normal tissue model (Fig. 1D; ref. 14). These studies demonstrated that the sensitization to doxorubicin, etoposide, and panobinostat observed in the 4T1 cells did not extend to the ESC (Fig. 1E). We did observe sensitization of BPTF KO ESCs to paclitaxel, which served as a rationale for not pursuing additional studies with this drug. Instead, we focused our follow-up studies on the TOP2 poisons doxorubicin and etoposide because these drugs are widely utilized (10), and have similar mechanisms of action (18).

Overexpression of human c-Myc-BPTF in 4T1 cells resulted in slightly reduced toxicity to the TOP2 poison, doxorubicin, supporting a role for BPTF in therapy resistance (Supplementary Fig. S1A and S1B). To test the possibility that drug accumulation and/or efflux might explain the drug sensitization observed with the BPTF KD 4T1 cells, we measured and observed equivalent cellular accumulation of doxorubicin (Supplementary Fig. S1C and S1D), as well as equivalent accumulation and efflux of pump substrates rhodamine 6G (19) and Hoechst 33342 (ref. 20; Supplementary Fig. S1E–S1J), in these cells suggesting that drug accumulation or efflux is not regulated by BPTF.

Additional studies demonstrated that WT and BPTF KD 4T1 cells showed equivalent sensitivity to the TOP1 poison, camptothecin (Supplementary Fig. S2A) and to the TOP2 catalytic inhibitor, ICRF-187 (Supplementary Fig. S2B), suggesting that BPTF suppresses toxicity selectively to TOP2 poisons. Unlike doxorubicin and etoposide, which trap TOP2cc, ICRF-187 inhibits TOP2 by blocking its ATP hydrolytic activity (21). Consequently, BPTF could regulate sensitivity to doxorubicin and etoposide, at least in part, by regulating the resolution of TOP2 crosslinking, enhancing DNA double-strand breaks (DSB). Consistent with this hypothesis, flow cytometric analysis showed increased DNA-damage marker, γH2AX (22), after exposure to doxorubicin and etoposide but not paclitaxel or radiation in BPTF KD cells (Fig. 2A); which was supported by quantitative Western blotting (Supplementary Fig. S2C–S2E). In further support for BPTF in regulating therapy induced DNA damage, human c-Myc-BPTF overexpression in 4T1 cells resulted in reduced γH2AX as measured by flow cytometry (Supplementary Fig. S2F).

Coincident to increased γH2AX abundance, we observed that BPTF KD resulted in increased doxorubicin- and etoposide-stabilized TOP2cc. Using an in vivo complex of enzyme (ICE) assay (15), enhancement of both alpha and beta TOP2cc levels was observed in doxorubicin- and etoposide-treated BPTF KD cells compared with control cells (Fig. 2B–D). Western blotting confirmed that the elevation in TOP2cc levels was not a result of BPTF-regulated TOP2 protein expression (Supplementary Fig. S2G). To determine whether the observed TOP2cc elevation in BPTF KD cells was a result of enhanced TOP2 chromatin attachment, we performed chromatin fractionation studies. We found that BPTF KD did not alter the levels of chromatin-bound TOP2α and TOP2β, both in the absence and presence of doxorubicin (Supplementary Fig. S2H), suggesting that BPTF does not affect TOP2 chromatin localization, and that the increase in TOP2cc levels in BPTF cells as observed in Fig. 2B–D, likely resulted from a defect in removing TOP2cc from the DNA in BPTF KD 4T1 cells. Consistent with BPTF's role in regulating doxorubicin-induced DNA damage, immunofluorescence imaging showed that BPTF is nuclear prior to and after doxorubicin treatment (Supplementary Fig. S2I). Taken together, these results support a model where BPTF resides in the nucleus to promote the repair of TOP2-linked DSBs.

An analysis of apoptosis by flow cytometric measurement of AnnexinV/7AAD staining (23) showed that BPTF KD only modestly altered the extent of doxorubicin-, etoposide-, and paclitaxel-induced apoptosis (Fig. 3A) and is unlikely to significantly contribute to the observed drug sensitization. Quantification of senescence by measuring SA-β-gal activity using C₁₂FDG (24) showed that doxorubicin, etoposide, and paclitaxel all induced cellular senescence (Fig. 3B). However, the overall extent of therapy-induced senescence was unaltered in the BPTF KD cells (Fig. 3B), indicating that sensitization is not a consequence of increased senescence.

It is well established that a primary response to anticancer chemotherapeutic drugs and radiation is autophagy (25). Flow cytometric and microscopy-based quantification of autophagy by acridine orange (AO) staining (26) shows a substantial increase in doxorubicin and etoposide-induced autophagy with BPTF KD (Fig. 3C; Supplementary Fig. S3A), whereas BPTF does not appear to regulate paclitaxel induced autophagy (Fig. 3C). Overexpression of human c-Myc-BPTF in 4T1 results in decreased doxorubicin-induced autophagy (AO staining), further supporting its roles in regulating therapy induced autophagy (Supplementary Fig. S3B). In addition to the increase in AO staining, the use of fluorescent dyes that selectively label lysosomes and mitochondria shows a substantial increase in co-localization in BPTF KD cells treated with doxorubicin and etoposide but not paclitaxel (Fig. 3D; Supplementary Fig. S3C).

The colocalization of mitochondria with lysosomes is characteristic of mitophagy, an autophagy-promoted process that selectively targets defective mitochondria (27). Another critical measure of autophagy is the conversion of microtubule-associated proteins 1A/1B light chain 3B (hereafter called LC3B) from LC3B-I to LC3B-II (28). Western blot analysis of LC3B demonstrated increased LC3B conversion (Ratio LC3B-II/LC3B-I) in the 4T1 BPTF KD cells prior to doxorubicin treatment (Fig. 3E). These results are consistent with an overall increase in autophagy, including the selective process of mitophagy, in the BPTF KD 4T1 cells.

To interrogate the potential involvement of autophagy in the sensitization of breast cancer cells to doxorubicin by BPTF KD, we depleted ATG5 by shRNA in the WT and BPTF KD 4T1 cell lines (Supplementary Fig. S3D). Evaluation of LC3B conversion confirmed that ATG5 KD compromised the capacity of 4T1 cells to undergo autophagy, as measured by a decreased ratio of LC3-II to LC3-I (Fig. 3E). Using in vitro clonogenic survival assays, we further determined that the autophagy induced by doxorubicin was nonprotective in the 4T1 cells in vitro, as reported previously (29). This conclusion is based on the observation that ATG5 KD failed to further sensitize the WT and BPTF KD 4T1 cells to doxorubicin exposure (Fig. 3F).

To determine whether the chemosensitization observed in cell culture by BPTF KD would also be observed in vivo, we surgically implanted WT and BPTF KD 4T1 cells into the fourth mammary fat pad of syngeneic BALB/c mice. Consistent with previous results (2), we did not observe any significant differences in BPTF KD 4T1 tumor growth in untreated mice (Fig. 4A). The tumor-bearing mice received three administrations of 5 mg/kg of doxorubicin at 7-day intervals, a dose that has minimal effects on the growth of WT 4T1 breast tumors (Fig. 4B). However, the doxorubicin treatments resulted in significant reductions in BPTF KD 4T1 tumor volume over the entire time period (Fig. 4C). These results replicate the cell culture findings demonstrating that BPTF KD 4T1 cells are more sensitive to doxorubicin and, by extension, that BPTF suppresses the antitumor effects of doxorubicin.

In mice, WT or BPTF KD 4T1 cells in combination with ATG5 KD grew similarly as tumors (Fig. 4D). In contrast to the clonogenic survival assays, doxorubicin treatment of mice bearing WT 4T1 tumors with ATG5 KD demonstrated that the sensitization to doxorubicin was modestly improved when autophagy was inhibited, demonstrating that autophagy is modestly protective to WT 4T1 tumors in vivo (Fig. 4E). Strikingly, the improvement in tumor sensitization to doxorubicin with BPTF KD was largely lost when combined with ATG5 KD (Fig. 4F). These results suggest that, in contrast to cell culture, functional autophagy is required for sensitization by BPTF KD in vivo.

To investigate the clinical significance of the above findings, we tested whether BPTF is a druggable target using a specific inhibitor of the BPTF bromodomain identified previously as AU1 (Fig. 5A and B; ref. 30). Pretreatment of 4T1 cells with AU1 resulted in increased sensitivity to doxorubicin (Fig. 5C) but not radiation (Supplementary Fig. S4A), as was the case for the BPTF KD cells (Fig. 1C). Interestingly, AU1 treatment did not recapitulate the sensitization to paclitaxel (Supplementary Fig. S4B) observed for BPTF KD cells (Fig. 1C), suggesting that AU1, and inhibition of the bromodomain, may not affect all of BPTF's functions.

Using BPTF KD2 because it was the most efficient KD (Fig. 1B), we further demonstrate that AU1 treatment of BPTF KD 4T1 cells did not further sensitize the cells to doxorubicin, which is consistent with AU1 inhibiting BPTF (Fig. 5D). In addition, the combination of AU1 and doxorubicin resulted in increased DNA damage as determined by flow cytometric measurement of γH2AX staining (Fig. 5E), increased autophagy (AO staining; Fig. 5F) and doxorubicin and etoposide-trapped induced TOP2cc by ICE (Supplementary Fig. S4C; Fig. 5G). AU1 treatment of BPTF KD cells did not further increase doxorubicin- and etoposide-induced TOP2cc, consistent with similar outcomes with BPTF KD (Fig. 2B–D), further suggesting that AU1 has specificity for BPTF (Supplementary Fig. S4C; Fig. 5G).

Consistent with our observations using BPTF KD 4T1 cells in vitro (Fig. 3F), ATG5 KD in combination with AU1 treatment did not alter the sensitivity of 4T1 cells to doxorubicin (Fig. 5H). However, in vivo treatment of established WT 4T1 tumors with AU1 resulted in a significant reduction of tumor weight when combined with doxorubicin, which confirms our previous experiments using genetic BPTF KD (Fig. 5I and J). Intratumoral injections of AU1 were used for these studies because AU1 has poor pharmacokinetics (31). Identical experiments using ATG5 KD tumors revealed reduced sensitization to doxorubicin with AU1 treatments, supporting a role for autophagy in the therapeutic benefits from BPTF inhibition (Fig. 5K and L). Histologic analysis of tumors treated with combinations of doxorubicin and AU1 showed an increase in the necrotic portion of the tumor, which stained positive for cleaved caspase-3 (Supplementary Fig. S4D). A further analysis of the nonnecrotic portions of the tumors also showed an increase in cleaved caspase-3 positive cells with the combined treatment (Fig. 5M). These enhancements of cleaved caspase-3 with combined AU1 and doxorubicin treatment were not observed in the ATG5 KD tumors (Fig. 5M), which is consistent with the dependence of sensitization by BPTF KD on functional autophagy shown in Fig. 4.

To determine whether NURF suppresses the sensitivity of human TNBC to chemotherapeutic agents, we generated control and BPTF KD MDA-MB-231 cells by lentivirus transduction (Fig. 6A). As with 4T1 cells, BPTF KD MDA-MB-231 cells showed increased sensitivity to doxorubicin, etoposide, and paclitaxel (Fig. 6B). Like 4T1 tumors, the sensitivity to doxorubicin was enhanced in BPTF KD MDA-MB-231 tumor–bearing mice, and this increased sensitization correlated with elevated therapy induced apoptosis, as measured by cleaved caspase-3 (Supplementary Fig. S5A–S5C). Also, as was the case with the mouse 4T1 TNBC model, AU1 treatment of MDA-MB-231 cells resulted in enhanced sensitivity to doxorubicin, but not paclitaxel (Fig. 6C). The ability of AU1 to recapitulate the sensitization of TNBC specifically to doxorubicin but not paclitaxel was conserved. As was the case in the mouse 4T1 model, flow cytometric analysis of γH2AX staining indicated that BPTF significantly suppresses doxorubicin-induced DNA damage in MDA-MB-231 cells (Fig. 6D), and also doxorubicin-induced mitophagy, as suggested by increased overlap of lysosomes and mitochondria (Fig. 6E). As published previously, autophagy is protective in the WT MDA-MB-231 cells (comparing WT and WT/ATG5 KD cells treated with doxorubicin; Supplementary Fig. S5D; ref. 32), in contrast to being nonprotective in the 4T1 cells (Fig. 3F). Autophagy appeared to be required for the sensitization of AU1-treated MDA-MB-231 cells to doxorubicin (Fig. 6F), as sensitization from AU1 and doxorubicin treatment was abrogated in ATG5 KD cells. As with AU1 treatment, sensitization to doxorubicin was also observed in BPTF KD MDA-MB-231 cells, and this sensitization required ATG5 (Fig. 6G and H). The requirement for autophagy in the sensitization of BPTF KD MDA-MB-231 cells to doxorubicin was also observed after the cells were aggregated into spheroids (Fig. 6I). Annexin V/7AAD staining demonstrated that apoptosis was increased in doxorubicin-treated BPTF KD MDA-MB-231 cells as compared with WT MDA-MB-231 cells, and this increase was reduced with ATG5 KD (Fig. 6J). Taken together, these results demonstrate that inhibition of BPTF can improve the response of breast cancer cells to TOP2-targeted chemotherapeutic agents and implicate autophagy in sensitization.

To assess whether BPTF more broadly regulates therapy resistance in breast cancer, we tested the metastatic TNBC 66cl4 (33), the uncharacterized metastatic 4T07 (34), and ER-positive 67NR cell lines (35) for doxorubicin sensitization after AU1 pretreatment by measuring clonogenic survival (Supplementary Fig. S5E). From these experiments, we observed that AU1 treatment increased sensitivity to doxorubicin in both 66cl4 and 4T07, but not the ER-positive and nonmetastatic 67NR cell line. The enhancements in sensitivity by AU1 to doxorubicin in 66cl4 and 4T07 cells coincided with increased DNA damage (γH2AX) and autophagy (AO staining; Supplementary Fig. S5F and S5G). In addition, AU1 and doxorubicin treatment in 66cl4 and 4T07 cells (using MDA-MB-231 as a positive control) coincided with increased TOP2α crosslinking over doxorubicin alone as measured by ICE (Supplementary Fig. S5H). To determine whether the AU1 and doxorubicin combination has toxicity to other ER-positive cell lines, we repeated the assays using the ER-positive MCF7 cell line (36). From these clonogenic survival experiments, we observed significant toxicity to the MCF7 cells for the combination treatment; however, this was in the context of significant toxicity for AU1 and doxorubicin, each alone, suggesting that the effects of the combination treatment were likely simply additive (Supplementary Fig. S5I). From these results it appears that AU1 treatment can sensitize additional TNBC and metastatic breast cell lines to the toxic effects of doxorubicin, and that this sensitization is not readily observed to ER-positive breast cancer cell lines.

Epigenetic changes to the cancer genome that promote cancer-specific biology are reversible and, as such, have been pursued as therapeutic targets (1). In contrast to DNMTs and HDACs, which have been targeted for decades, inhibitors of chromatin remodeling enzymes are just being developed (37). In this study, we aimed to determine how the chromatin remodeling complex NURF functions to regulate the sensitivity of breast cancer cells to chemotherapeutic drugs and radiation. By targeting BPTF, the unique and essential subunit of NURF (17), we show that NURF selectively suppresses the effects of chromatin-targeted chemotherapies including doxorubicin, etoposide, and panobinostat. These results are consistent with independent studies in tissue culture with liver cancer models showing that BPTF KD sensitizes to cisplatin and 5-fluorouracil (3) and to HDAC inhibitors (7). The results are also consistent with similar studies with the NuRD remodeling complex and daunorubicin, cytarabine (38), and docetaxel (39), and SWI/SNF, which suppresses sensitivity to a large number of chemotherapies (40). Hence, there is growing evidence that inhibiting chromatin remodeling complexes sensitizes multiple cancer types to broad classes of chemotherapeutic agents.

One important mechanism by which BPTF suppresses doxorubicin and etoposide toxicity is inhibition of TOP2cc. These mechanisms are distinct from BPTF suppression of HDACi activity, which does not operate as a TOP2 poison, and is therefore a topic for another study. Both BPTF and TOP2 localize to regions of open chromatin including promoters, enhancers, and CTCF-binding sites, making their direct functional interaction plausible (41, 42). Equivalent TOP2 chromatin association after doxorubicin exposure with BPTF KD suggests it is unlikely that BPTF regulates access of TOP2 to DNA. The elevated levels of TOP2cc suggest that BPTF participates in the repair of trapped TOP2cc, possibly including TDP2, which has specific functions in resolving TOP2cc (43). It is unlikely that BPTF regulates general DSB repair pathways, as radiation- and doxorubicin-induced DNA damage use the Mre11 repair pathways and the defects in DNA damage repair were selectively observed with doxorubicin-induced DNA damage and not to radiation-treated cells.

An unanticipated outcome from this study is that the sensitization of BPTF KD cells to chemotherapeutic agents requires autophagy. Despite the induction of both senescence and apoptosis with doxorubicin-induced DNA damage, BPTF predominantly affects autophagy as shown by increased AO staining, mitophagy, and LC3B conversion. This could be the result of BPTF regulating autophagy control points including Ulk1 (44) and/or Bcl2/Beclin1 (45), or by regulating the expression of autophagy components or lysosome biogenesis (46). The latter is plausible as many autophagy and lysosome biogenesis genes are regulated by epigenetic regulators (reviewed extensively in ref. 46). Changes in the autophagy pathway may not necessarily influence cell autonomous toxicity but could selectively contribute to the loss of protective autophagy observed in tumors (a non-cell–autonomous effect) as we observed for the 4T1 tumors. Possible contributions to in vivo tumor growth control could involve changes in autophagy-regulated angiogenesis (47), sensitization to therapies in a hypoxic tumor environment (48), a therapy-induced antitumor immune response through inflammatory cytokines (49), or by inducing autophagic cell death (50).

The BPTF bromodomain inhibitor AU1 has been shown previously to inhibit the growth of the K562 human cancer cells in culture (31). This report extends these observations by showing that AU1 has anticancer activity when used in combination with TOP2 inhibitors using in vitro and in vivo tumor models. These results are different from those observed when using the SWI/SNF bromodomain inhibitor ADAADiN, which functions as a broad sensitizer to chemotherapeutic agents by suppressing ABC transporter expression (40). Unexpectedly, the BPTF inhibitor AU1 does not sensitize 4T1 or MBA-MB-231 cells to paclitaxel, for which we showed sensitization for BPTF KD. These results suggest that specifically inhibiting the BPTF bromodomain, the target for AU1 (30), may only inhibit a subset of BPTFs function in sensitization to chemotherapeutic agents. The regulation of BPTF sensitization to paclitaxel could occur through other functional domains in BPTF, including the C-terminal PHD finger, which is well known to interact with H3K4me3, the N-terminal PHD finger, or through interaction surfaces important for transcription factor binding (17).

No disclosures were reported.

L. Tyutyunyk-Massey: Formal analysis, investigation. Y. Sun: Formal analysis, investigation, writing–review and editing. N. Dao: Investigation. H. Ngo: Investigation. M. Dammalapati: Investigation. A. Vaidyanathan: Investigation. M. Singh: Investigation. S. Haqqani: Investigation. J. Haueis: Investigation. R. Finnegan: Investigation. X. Deng: Formal analysis. S.E. Kirberger: Resources. P.D. Bos: Formal analysis, visualization. D. Bandyopadhyay: Formal analysis. W.C.K. Pomerantz: Resources. Y. Pommier: Supervision, methodology, writing–review and editing. D.A. Gewirtz: Supervision, writing–review and editing. J.W. Landry: Conceptualization, formal analysis, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing.

The authors thank Virginia Moose Association, Massey Cancer Center, VCU SOM Bridge Funding, and the Department of Defense grant W81XWH1910489 (to J.W. Landry and D.A. Gewirtz), the National Institute of General Medical Sciences R01GM121414 (W.C.K. Pomerantz), the Intramural Program of the NCI (Z01-BC-006161; to Y. Sun and Y. Pommier), and support from the NIH-NCI Cancer Center Support Grant P30 CA016059.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.