The bromodomain inhibitor PFI-3 sensitizes cancer cells to DNA damage by targeting SWI/SNF

Daye Lee1, Da-Yeon Lee1, You-Son Hwang1, Hye-Ran Seo, Shin-Ai Lee, and Jongbum Kwon

Department of Life Science

The Research Center for Cellular Homeostasis Ewha Womans University
Seoul, Korea 120-750

Corresponding author: Jongbum Kwon

Science Building C-308, Ewha Womans University, 52 Ewhayeodae-gil, Seodaemun-gu,
Seoul 03760 Korea.

Phone: 82-2-3277-4334

Fax: 82-2-3277-3760

Email: [email protected]

1 These authors contributed equally to this work.

Running title: PFI-3 sensitizes cancer cells to DNA damage

Conflict of interest: The authors declare no competing financial interests.


Many chemotherapeutic drugs produce double-strand breaks (DSBs) on cancer cell DNA, thereby inducing cell death. However, the DNA damage response (DDR) enables cancer cells to overcome DNA damage and escape cell death, often leading to therapeutic resistance and unsuccessful outcomes. It is therefore important to develop inhibitors that target DDR proteins to render cancer cells hypersensitive to DNA damage. Here, we investigated the applicability of PFI-3, a recently developed bromodomain (BRD) inhibitor specifically targeting the SWI/SNF chromatin remodeler that functions to promote DSB repair, in cancer treatment. We verified that PFI-3 effectively blocks chromatin binding of its target BRDs and dissociates the corresponding SWI/SNF proteins from chromatin. We then found that, while having little toxicity as a single agent, PFI-3 synergistically sensitizes several human cancer cell lines to DNA damage induced by chemotherapeutic drugs such as doxorubicin. This PFI- 3 activity occurs only for the cancer cells that require SWI/SNF for DNA repair. Our mechanism studies show that PFI-3 exerts the DNA damage-sensitizing effect by directly blocking SWI/SNF’s chromatin binding, which leads to defects in DSB repair and aberrations in damage checkpoints, eventually resulting in increase of cell death primarily via necrosis and senescence. This work therefore demonstrates the activity of PFI-3 to sensitize cancer cells to DNA damage and its mechanism of action via SWI/SNF targeting, providing an experimental rationale for developing PFI-3 as a sensitizing agent in cancer chemotherapy.


This study, revealing the activity of PFI-3 to sensitize cancer cells to chemotherapeutic drugs, provides an experimental rationale for developing this BRD inhibitor as a sensitizing agent in cancer chemotherapy.


Chemo- and radiotherapy are the most widely used treatment modalities for cancer. Many chemotherapeutic drugs, such as doxorubicin, and ionizing radiation (IR) work by generating double-strand breaks (DSBs) on cancer cell DNA and thereby inducing cell death. However, the DNA damage response (DDR), a complex network of cellular pathways that sense, signal and repair DNA lesions, enables tumor cells to overcome DNA damage and thereby escape cell death, which often leads to therapeutic resistance and unsuccessful outcome in cancer treatment. Targeting DDR proteins can be a useful strategy to sensitize tumor cells to DNA damage and thereby increase the efficacy of cancer treatment. Therefore, it is important to discover DDR proteins whose inactivation renders cancer cells hypersensitive to DNA damage and develop inhibitors that specifically target these proteins (1-4).SWI/SNF chromatin-remodeling complex functions in multiple cellular processes, including DNA repair (5). The mammalian SWI/SNF complexes (denoted SWI/SNF hereafter and also known as BAF) contain either BRG1 or BRM as a catalytic ATPase, which confers the complex with remodeling activity, and at least ten non-catalytic subunits, including BAF180 (PBRM1), which play regulatory role. SWI/SNF exists in many different forms in the subunit compositions and is classified into two major groups, BAF and PBAF. While BAF contains either BRG1 or BRM as an ATPase, PBAF always contain BRG1 as an ATPase and BAF180 as an associating factor (6,7). Both BAF and PBAF have direct role in DSB repair. Depletion or inactivation of SWI/SNF subunits, including BRG1 and BRM, reduces DSB repair and sensitizes cells to DSB-inducing agents, and SWI/SNF rapidly localizes at DSB-surrounding chromatin (8-16). We have previously shown that BRG1 lacking bromodomain (BRD), an approximately 110 amino acid protein domain that recognizes and binds acetyl-lysine motifs, is unable to stimulate DSB repair and BRG1 BRD
3exerts dominant negative activity against BRG1 to inhibit DSB repair, indicating that SWI/SNF binding to chromatin via BRD is important for DSB repair (17,18). The follow-up work showed that blocking BRG1 chromatin binding by overexpression of BRG1 BRD inhibits DSB repair and sensitizes cancer cells to DSB-generating agents, such as doxorubicin and IR. This proof-of-concept study verified that BRG1 BRD is a target to enhance the efficacy of cancer chemo- and radiotherapy (19).

The human genome encodes 61 BRDs found in 42 proteins and many of these proteins are chromatin regulators, such as chromatin remodeler and histone modifying enzyme, which function in a wide array of biological processes, including DNA repair (20). BRDs are classified into eight families based on similarity of their sequence and structure. Despite of their sequence diversity, all BRDs share a conserved fold comprising left-handed four helix bundles linked by loops with variable lengths that form a central deep and narrow hydrophobic cavity, which enables this protein domain to recognize the acetyl-lysine motifs in a sequence-dependent manner. This feature of the BRD fold has made it possible to develop specific small-molecule inhibitors against many different BRDs (21-24). The first examples of such inhibitors were JQ1 and I-BET, which target the bromo- and extra-terminal domain (BET) family proteins and exhibited anti-cancer and anti-inflammatory activities (25,26). Subsequently, a number of small-molecule inhibitors targeting various other BRDs were developed (27-29). One of those inhibitors is PFI-3. In vitro experiments showed that PFI-3 specifically binds to each BRD of BRG1 and BRM and selectively to the fifth BRD among the six BRDs of BAF180, all belonging to the family VIII of BRDs (30,31).In this study, we investigated whether PFI-3 inhibits SWI/SNF and sensitizes cancer cells to DNA damage to test its applicability in cancer treatment. We found that, while exhibiting little cytotoxic activity by itself, PFI-3 synergistically sensitizes several humancancer cells to DNA damage induced by the chemotherapeutic drugs, such as doxorubicin. Our data suggest that PFI-3 exerts this activity by direct blockade of SWI/SNF’s chromatin binding, which leads to defective DNA repair and aberrant damage checkpoint, ultimately resulting in increase of cell death primarily via necrosis and senescence.

Materials and Methods

Cells and plasmids

A549, HT29, H460, H1299 and U2OS cells were purchased from ATCC (Manassas, VA) and cultured according to the vendor’s instructions. The cell lines were authenticated with DNA fingerprinting using STR (short tandem repeat) markers every 50 passages and were tested for the absence of Mycoplasma contamination using the e-Myco VALiD Mycoplasma PCR Detection Kit (Intron Biotechnology). The U2OS stable cells were generated by transfection with the vector expressing GFP-conjugated dimeric form of BRG1 BRD (19) followed by selection with G418 at 800 mg/ml. The vectors expressing GFP-conjugated monomeric BRG1 BRD or BRM BRD were previously described (32).
Inhibitors and drugs

PFI-3 and BI-9564 was obtained from the Structural Genomics Consortium ( and Tocris (5072, UK). Doxorubicin and etoposide were purchasedfrom Sigma-Aldrich, and the BRG1/BRM-specific ATPase inhibitor from Chem Scene (NJ, USA). The chemicals were dissolved in DMSO, kept in aliquots at -20℃, and thawed immediately before use for experiments.
Cell viability assayClonogenic assay was performed as previously described (17). After stained with 0.5% crystal violet, colonies were dissolved in 10% acetic acid before measuring absorbance at 5905nm with a spectraMax i3X plate reader (I3X-SC-ACAD, Molecular Devices). MTS assay was performed using the CellTilter 96 AQueous One Solution Cell Proliferation assay kit (Promega), with absorbance at 490 nm measured by the above-mentioned plate reader.

Chromatin fractionation assay

Biochemical fractionation was performed as previously described (33). Briefly, cells were incubated in the fractionation buffer (50 mM HEPES (pH 7.5), 150 mM NaCl, 1 mM EDTA, 0.5% NP-40, 0.5 mM PMSF, 5 µg/ml aprotinin, 5 µg/ml leupeptin, 5 µg/ml pepstatin A, and 10 mM NaF) on ice for 1 h, and then centrifuged at 16000g for 20 min to separate supernatant (chromatin-unbound fraction) from pellet (chromatin-bound fraction).
Statistical analysis

The significance of differences between measurements was evaluated by Student’s t-test in Microsoft Excel. A p value < 0.05 was deemed to indicate statistical significance.
Combination index (CI) was calculated according to the Chou-Talalay algorithm with the CompuSyn software and drug interaction was described as previously reported (34).
Additional materials and methodsIn situ cell extraction, comet assay, annexin-V staining, immunoblotting, transfection, RNA- Seq analysis, and the sequences of siRNAs and the sources of antibodies were provided in Supplementary data.


PFI-3 inhibits chromatin binding of its target BRDs and the corresponding SWI/SNF proteins within cells
To verify the activity of PFI-3 (Fig. 1A) within the cells, we first examined its ability to inhibit chromatin binding of BRG1 BRD by in situ cell extraction using U2OS cells stably
6expressing a dimeric form of BRG1 BRD linked to GFP and Myc, which has higher chromatin affinity than a monomeric form (19). After treatment of the cells with PFI-3 along with trichostatin A (TSA, to enhance assay window) for 2 h, we detergent-extracted unbound BRDs and determined the GFP intensity on the cells by fluorescence microscopy (Fig. 1B). PFI-3 treatment at 30- and 50-mM reduced the GFP intensity in a dose dependent manner, whereas the same concentrations of BI-9564, a specific inhibitor against BRDs of the BRD7 and BRD9 subunits of SWI/SNF (35), showed no effect (Fig. 1C), indicating that PFI-3 inhibited the chromatin binding of BRG1 BRD specifically. Then, we confirmed these results by chromatin fractionation. PFI-3 at 30 mM, but not BI-9564, significantly inhibited the chromatin binding of BRG1 BRD regardless of TSA treatment (Fig. 1D, S1A and B). PFI-3 inhibited the chromatin binding of BRG1 BRD as well as BRM BRD in a dose-dependent manner (Fig. 1E and F). We also confirmed that PFI-3 inhibited chromatin binding of the fifth BRD but not the other five BRDs of BAF180 (Fig. 1G).

Next, we determined whether PFI-3 dissociates endogenous SWI/SNF proteins from chromatin. Treatment of U2OS cells with PFI-3 at 50 mM for 2 h dissociated BRG1, BRM and BAF180 from chromatin regardless of TSA treatment whereas BI-9564 showed no effect (Fig. 1H and S1C). This PFI-3 activity was dose dependent (Fig. 1I). PFI-3 appears to be highly stable within the cells since its activity was still effective after treatment for 24 h (Fig. S1D). We confirmed these results by in situ detergent extraction combined with immunofluorescence microscopy. When the cells were treated with increasing PFI-3 up to 100 mM, the fluorescent signals specific for BRG1 (Fig. 1J) or BRM (Fig. 1K) proportionally decreased with the signals barely detected at 50-mM and higher concentrations, indicating that PFI-3 dissociated BRG1 and BRM from chromatin effectively and in a dose-dependent manner. As a negative control, BI-9564 did not dissociate BRG1 from chromatin even at 50-
7mM and higher concentrations (Fig. 1J). These results collectively demonstrated that PFI-3 inhibits chromatin binding of its target BRDs and the corresponding endogenous SWI/SNF proteins at pharmacologically significant concentrations.

PFI-3 sensitizes several human cancer cells to DNA damage.

Given that PFI-3 inhibits SWI/SNF, we investigated whether it sensitizes cancer cells to DNA damage. For this study, we chose A549 (lung cancer) and HT29 (colon cancer) as model cells representing not only different cancer types but also distinct SWI/SNF gene mutations; the former is BRG1-deficient and the latter BRG1-proficient while both express BRM and BAF180. First, we confirmed that PFI-3 dissociates SWI/SNF from chromatin in these cancer cells by chromatin fractionation (Fig. S2A and B) and in situ detergent extraction (Fig. S2C and D). We then found that, while having little cytotoxic activity by itself, PFI-3 sensitized both A549 and HT29 cells to doxorubicin and etoposide in certain combinations of drug concentrations in the assays to determine short-term viability (Fig. S3A and B) and long-term clonogenic ability (Fig. 2A and B). Calculation of the combination index indicated that the PFI-3 effects on the cancer cells’ sensitivity to doxorubicin and etoposide were synergistic (Fig. 2A and B). A similar degree of doxorubicin sensitization was observed for both cells whether a same amount of PFI-3 was treated multiple times with wash (Fig. S3C-E) or divided into small portions and given several times without wash (Fig. S3F-H), suggesting that stability was not an issue for the PFI-3 activity. PFI-3 exhibited a synergy with IR to kill A549 but this effect was observed only at a very high concentration (120 mM), and this high concentration of PFI-3 had some cytotoxic activity on the cells by itself (Fig. S3I). However, PFI-3 at 120 mM showed no combined effect with IR on HT29 cells, which possibly could be due to its own cytotoxic activity relatively stronger on this cell type compared to A549 (Fig. S3J). Then, we analyzed two more lung cancer cell lines, H460 and H1299, with similar
8SWI/SNF gene expression as HT29 and A549, respectively (Fig. 2C). PFI-3 exhibited the similar activity on H460 as A549 and HT29 cells; a synergy with doxorubicin but no cytotoxicity by itself (Fig. 2D). Interestingly, however, PFI-3 alone showed significant cytotoxicity on H1299 cells but no combined effect with doxorubicin (Fig. 2E). This difference was attributed to the fact that SWI/SNF is dispensable for DNA repair in H1299 cells (see below). All these results show that PFI-3 sensitizes cancer cells to DNA damage and this activity of PFI-3 is not general but rather depends on both cancer cell type and DNA damaging agent.

A recent study reported that inhibiting ATPase activity is more appropriate than targeting BRD for inducing a synthetic lethality of SWI/SNF-mutant cancer cells (36). We therefore compared the two strategies for cancer cell cytotoxicity in combination with doxorubicin. The recently developed dual BRG1 and BRM ATPase inhibitor (ATPasei) (37) had a strong cytotoxic activity by itself on A549 and HT29 cells with the IC50 values of 0.50 and 1.49 mM, respectively (Fig. S3K). When treated in combination with doxorubicin, the ATPasei exhibited only an additive effect of cytotoxicity on A549 cells in contrast to PFI-3 showing a synergy as observed before (Fig. 2F). The reason the ATPasei has no synergy with doxorubicin is likely because of its own strong cytotoxic activity. Therefore, while inhibiting the ATPase activity appears to be far more potent in treatment of SWI/SNF mutant cancers, targeting BRD could have an advantage of synergy effect in combination treatment with doxorubicin.PFI-3 sensitizes cancer cells to DNA damage by inhibiting SWI/SNF-driven DNA repair. Next, we investigated whether PFI-3 sensitizes the cancer cells to DNA damage by inhibiting DNA repair. First, we asked whether SWI/SNF functions in DNA repair in the cancer cells.
As assessed by neutral comet assay detecting DSBs specifically, siRNA knockdown of either9BRM or BRM/BAF180 led to a large defect in DSB repair after doxorubicin treatment in A549 cells (Fig. 3A and B), showing that SWI/SNF is critical for DSB repair in this cell type. Then, we determined the effects of PFI-3 on DNA repair. Notably, while not inducing DSBs by itself, PFI-3 largely inhibited the repair of doxorubicin-induced DSBs that would have otherwise efficiently proceeded with recovery time (Fig. 3C). We confirmed these results by the observations that g-H2AX, the physiological DSB indicator (38), after induction by doxorubicin decreased with recovery time under normal conditions but this decrease was attenuated by PFI-3 as assessed by immunoblotting (Fig. 3D) and immunofluorescence microscopy detecting foci formation (Fig. 3E and F). Importantly, depletion of BRM or BRM/BAF180 canceled the doxorubicin sensitizing activity of PFI-3 (Fig. 3G and H), suggesting that PFI-3 exerts this activity by targeting SWI/SNF.

Then, we analyzed the other three cell lines. siRNA knockdown of BRG1/BRM or BRG1/BRM/BAF180 showed that SWI/SNF is important for the repair of doxorubicin- induced DSBs in HT29 (Fig. S4A and B) and H460 cells (Fig. S4C and D). PFI-3 inhibited DSB repair after doxorubicin treatment in both HT29 (Fig. 3I) and H460 cells (Fig. S4E).Interestingly, the kinetics of DNA repair in these two cell lines was different from A549 in that unrepaired DSBs decreased shortly after doxorubicin treatment and then continued to increase until the end of the time course analyzed (24 h), which may be due to the differences in BRG1 expression and/or other genetic factors. Notably, depletion of BRM or BRM/BAF180 did not affect DSB repair after doxorubicin treatment in H1299 cells (Fig. S4F and G), indicating that SWI/SNF is dispensable for DSB repair in this cell line. Consistently, PFI-3 exhibited no effect on DSB repair after doxorubicin treatment (Fig. S4H). These results explain why PFI-3 had no DNA damage sensitizing activity on H1299 cells. Thus, the reduced viability of H1299 by PFI-3 alone is likely attributed to the SWI/SNF functions that10are essential for cell survival but unrelated to DNA repair. Taking all these findings together, we concluded that PFI-3 sensitize cancer cells to DNA damage by inhibiting SWI/SNF- driven DNA repair.

PFI-3 blocks DNA damage-induced SWI/SNF binding to chromatin.

Next, we investigated the mechanism by which PFI-3 inhibits the SWI/SNF-driven DNA repair. First, we examined how SWI/SNF responds to DNA damage by determining its chromatin binding various times after doxorubicin treatment. Chromatin fractionation showed that BRM/BAF180 and BRG1/BRM rapidly bound to chromatin after doxorubicin treatment in A549 and HT29 cells, respectively, and this binding increased continuously until the end of the time course analyzed (24 h) (Fig. 4A and B). Therefore, although the repair kinetics are different between A549 and HT29 cells, a continuous chromatin binding of SWI/SNF seems to be important for the proper responses to doxorubicin damage in both cells. Then, we determined whether PFI-3 inhibits the DNA damage-induced SWI/SNF’s chromatin binding by chromatin fractionation. Notably, the chromatin binding of BRM/BAF180 in A549 (Fig.4C and S5A) and BRG1/BRM in HT29 cells (Fig. 4D and S5B) increased after doxorubicin treatment, which was abolished by PFI-3. We confirmed these findings by in situ detergent extraction, which showed that PFI-3 abolished the increase of doxorubicin-induced chromatin binding of BRM in A549 (Fig. 4E) and BRG1/BRM in HT29 cells (Fig. 4F). These results suggest that PFI-3 inhibits the repair of doxorubicin-induced DNA damage by blocking SWI/SNF chromatin binding.PFI-3 affects DNA damage checkpoint after doxorubicin treatment.

Given the DNA repair inhibition by PFI-3, we determined its impacts on the cell-cycle checkpoint by recovering A549 and HT29 cells from doxorubicin damage with or without PFI-3 for various times up to 48 h (Fig. 5A). While both cells exhibited G2/M arrest after11doxorubicin treatment, PFI-3 treatment led to some delay in the establishment of G2/M arrest in A549 cells and a strengthened and persistent G2/M arrest in HT29 cells (Fig. 5B and S6A). Analysis of several checkpoint proteins, such as Chk1/2, Cdk1 and p53, showed that, while doxorubicin treatment even in the presence of PFI-3 activated the damage checkpoint, PFI-3 had significant impacts on the checkpoint activity in both cells but in somewhat different ways. In A549 cells, the activity of Chk1, but not Chk2, was significantly attenuated by PFI-3, as assessed by their activating phosphorylation at Ser-317/345 (39). The inhibitory phosphorylation of Tyr-15 on Cdk1, the master kinase that regulates the transition from G2 to mitosis (40), was largely defective in the presence of PFI-3. In contrast, the p53 activation was significantly enhanced by PFI-3, as determined by Ser-20 phosphorylation (41). In HT29 cells, PFI-3 rather increased the Chk1 activity after doxorubicin induction and exhibited little effect on the activity of the other checkpoint proteins (Fig. 5C and S6B). PFI-3 itself had no effect on the cell-cycle profile or the damage checkpoint (Fig. 5B and C). Therefore, it appears that PFI-3 induces some aberrations in the doxorubicin-induced damage checkpointin both A549 and HT29 cells, possibly via defective DNA repair, although its specific impacts on the damage checkpoint seem somewhat different between the two cell types. PFI-3 increases doxorubicin-induced cell necrosis and senescence.

Next, we investigated how PFI-3 decreases the cancer cell viability after doxorubicin treatment. Annexin-V staining showed that doxorubicin treatment induced cell death mostly via necrosis in both A549 and HT29 cells, which was increased by PFI-3 (Fig. 6A and B).Although its contribution to cell death was small, apoptosis after doxorubicin treatment was significantly increased by PFI-3 in HT29 cells but not A549 (Fig. 6A and B), which was confirmed by PARP cleavage (Fig. 6C). Although A549 and HT29 cells showed a similar degree of sensitivity to the combined treatment of doxorubicin and PFI-3 in clonogenic assay,

12the former exhibited much less necrosis and apoptosis compared to the latter. Notably, doxorubicin treatment increased senescence in A549 cells, which was largely enhanced by PFI-3 (Fig. 6D), suggesting that, in addition to necrosis, senescence also contributes significantly to the reduced viability of this cell after the combined treatment of doxorubicin and PFI-3. PFI-3 alone had no effect on the senescence (Fig. 6D). Therefore, we concluded that PFI-3 sensitizes A549 cells to doxorubicin by increasing both necrosis and senescence while it does so for HT29 primarily via necrosis and apoptosis to some extent.Impacts of PFI-3 on gene expression in cancer cells.

Finally, we investigated the impacts of PFI-3 on gene expression in A549 and HT29 cells. The cells after doxorubicin treatment were recovered with or without PFI-3 and analyzed for genome-wide transcriptome profiles by RNA sequencing. Total 456 and 341 genes were up- or downregulated by ≥ 2 folds relative to control (DMSO) by treatment of PFI-3, doxorubicin or both in A549 and HT29 cells, respectively (Fig. S7A and Table S1 and 2). Unbiased hierarchical clustering of each of those gene sets revealed that the samples of doxorubicin and PFI-3/doxorubicin were grouped closely together whereas DMSO was clustered with PFI-3 (Fig. S7A). This pattern became more obvious when clustering was performed for the genes that were categorized as associated with DDR, including DNA repair, cell cycle control, cell death and proliferation (151 genes in A549 and 42 in HT29) (Fig. 7A and B, Table S3 and 4). This clustering also revealed that, while PFI-3 by itself exhibited a similar pattern as control, it further increased and decreased doxorubicin-upregulated and -downregulated gene expression, respectively (Fig. 7A and B), possibly the results of the amplification of DDR signaling reinforced by defective DNA repair.Plotting Venn diagrams for each of the fc ≥ 2 genes in A549 and HT29 cells showed that, although many genes were regulated in common by PFI-3, doxorubicin and PFI-133/doxorubicin, far more genes were regulated independently (Fig. 7C and D). In A549 cells, for example, only 25 out of 159 genes (15.7%) were regulated in common by PFI-3 and doxorubicin, and the remaining 134 genes (84.3%) were regulated independently. Notably, although significant percentages of PFI-3- (27 genes out of 72, 37%) and doxorubicin- regulated genes (74 genes out of 112, 66%) were overlapped with PFI-3/doxorubicin- regulated genes, the majority of the PFI-3/doxorubicin-regulated genes (248 genes out of 334, 74%) were unique with no overlapping with the PFI-3- or doxorubicin-regulated genes (Fig. 7C and D). These results suggest that, while each having distinct cellular activities, PFI-3 and doxorubicin in combination impose different impacts from individual treatment, which may be reflected on their synergistic cytotoxicity on the cancer cells.

To determine whether and how well PFI-3-regulated genes overlap with SWI/SNF- regulated genes, we performed transcriptome profiling coupled with gene set enrichment analysis (GSEA) for the cancer cells treated with PFI-3 or knockdowned for SWI/SNF (knockdown for BRM or BRM/BAF180 in A549; BRG1/BRM or BRG1/BRM/BAF180 in HT29) (Fig. S7B, Table S5 and 6). For both of the up- and downregulated genes (fc ≥ 2), there was a highly significant correlation between PFI-3 treatment and SWI/SNF knockdown in A549 and HT29 cells (FDR q-values = 0.00 for both) (Fig. 7E and F), suggesting that PFI- 3 targets SWI/SNF specifically. Importantly, no DNA repair genes were found among the PFI-regulated genes, suggesting that PFI-3 inhibits DNA repair by directly targeting SWI/SNF rather than via gene expression.


Here, we investigated the applicability of PFI-3 as a sensitizer for cancer chemotherapy based on our previous proof-of-concept study validating SWI/SNF as an appropriate target (19). We14verified that PFI-3 inhibits chromatin binding of its target BRDs and displaces the corresponding endogenous SWI/SNF proteins from chromatin at pharmacologically significant concentrations. We then found that, while having little cytotoxicity by itself, PFI-3 synergistically sensitizes several human cancer cells to DNA damage induced by chemotherapeutic drugs, such as doxorubicin. Our data suggests that PFI-3 exerts this activity by blocking the damage-induced SWI/SNF binding to chromatin, which leads to defective DNA repair and aberrant damage checkpoint, eventually resulting in increase of cell death primarily via necrosis and senescence (Fig. 7G). This work, demonstrating the activity ofPFI-3 to sensitize cancer cells to DNA damage by targeting SWI/SNF and its action mechanism, provides a strong experimental rationale for developing PFI-3 as a sensitizing agent for cancer chemotherapy.

Several lines of experimental evidence support our conclusion that PFI-3 sensitizes cancer cells to DNA damage by direct blockade of DNA repair via SWI/SNF targeting. First, PFI-3 not only dissociates SWI/SNF from chromatin under normal conditions but also inhibits it from binding to chromatin in response to DNA damage. Second, SWI/SNF depletion abolishes the activity of PFI-3 to sensitize cancer cells to DNA damage. Third, either SWI/SNF depletion or PFI-3 treatment inhibits DNA damage repair in cancer cells. In addition, DNA damage sensitization by PFI-3 occurs in the cancer cells that require SWI/SNF for DNA repair but not the cancer cells in which SWI/SNF is dispensable for DNA repair. Finally, a genome-wide transcriptome analysis shows that PFI-3 has no effect on the expression of DNA repair genes and that PFI-3-regulated genes well overlap with SWI/SNF- regulated genes.
Our results show that the phenotypic outcomes of PFI-3 in cancer cells are complex. PFI-3 sensitizes A549, HT29 and H460 cells to DNA damage with little of its own cytotoxic15activity whereas it has cytotoxicity on H1299 cells by itself with no DNA damage sensitization. The former three cells, but not the latter, require SWI/SNF for DNA repair. Thus, how PFI-3 acts on cancer cells appears to depend on whether the cancer cells rely on SWI/SNF for their survival or DNA repair. In addition, PFI-3, albeit at high concentrations, sensitizes A549 cells, but not HT29, to IR whereas it sensitizes both cells to the chemotherapeutic drugs, such as doxorubicin and etoposide, suggesting that the PFI-3 activity of sensitizing cancer cells to DNA damage depends on DNA damage source as well as cancer cell type. Further, although we expected that BRG1-deficient cells would be more sensitive to PFI-3 than BRG1-proficient cells simply due to less PFI-3 targets, we observed no clear correlation between BRG1 deficiency and the cells’ sensitivity to PFI-3. Thus, the sensitivity of cancer cells to PFI-3 in DNA damage does not depend on whether or not thecancer cells possess both BRG1- and BRM-based SWI/SNF complexes. Instead, other factors, such as chromatin affinity of involved BRDs and/or role of SWI/SNF in DNA repair, are likely the determinants of the efficacy of PFI-3 in particular cancer cells. Although further study will be necessary to clarify the complexity of the phenotypic outcomes of PFI-3, it could be an advantage to develop PFI-3 as an anticancer agent for treatment of specificcancer using specific modality.

PFI-3 has been previously investigated for its cellular and biological activities (30,31,36). In the study reporting the development of PFI-3, the authors showed that PFI-3 displaces transfected GFP-BRM from chromatin in U2OS cells in a fluorescence recovery after photobleaching assay, and that PFI-3 mimics the effects of BRG1 depletion on stemness gene expression in embryonic stem cells, leading to deprivation of their stemness and deregulated lineage specification (30). Another study reported that, although PFI-3 inhibits chromatin binding of transfected BRM BRD, it does not displace endogenous BRM from

16chromatin in A549 cells in in situ cell extraction assay (36). Our results, showing that PFI-3 displaces the endogenous BRM from chromatin in in situ cell extraction and chromatin fractionation assays using U2OS and A549 cells, are in keeping with the former but not the latter study. Although not clearly understood, the discrepancy between the in situ extraction data of the latter study and ours could be attributed to different experimental conditions used. For example, we used much lower concentrations of detergent compared to that study, which might have increased the assay window. In addition, the latter study also showed that PFI-3 alone has no cytotoxic activity on any of the tested cancer cells, including A549, H1299 and H460 cells, which is contradictory to our results that, while having no effect on A549 and H460 cells, PFI-3 by itself has cytotoxicity on H1299 cells. One possible explanation for this discrepancy could be that the apparent BRM level in our H1299 cells is much lower than that in the H1299 cells used in that study, as judged by direct comparison of the BRM levels between H1299 and H460 cells (see Fig. 2C in this paper and Fig. 2A in (36)). The low BRM level in our H1299 may render the cells sensitive to PFI-3 alone.

The success in evolving potent and highly specific inhibitors for the BET family of BRDs and their entering into clinical trials has stimulated intensive research activities to develop inhibitors targeting other BRDs (42). PFI-3 was one of the fruits of such research efforts. Although our work presented the possibility of potential application of PFI-3 as a sensitizer for cancer chemotherapy, many issues remain to be solved. Most importantly, the effective concentrations of PFI-3 on cancer cells are considerably high relative to the BET BRD inhibitors, such as JQ1 and PFI-1 (25,43). It is thus worth to evolve structural analogs of PFI-3 as to have higher affinity to its target BRDs and better cytotoxic efficacy on cancer cells, for instance, by using a molecular modeling approach. It is also of great value to pursue investigation to search in a systematic way for more cancer cell types that are sensitive to17

PFI-3 alone and/or sensitized to DNA-damaging chemotherapeutic drugs and IR by PFI-3.18


This work was supported by grants 2015M2A2A7A01041767 (J. Kwon), 2018R1A2B2007128 (J. Kwon) and 2019R1A5A6099645 (J. Kwon) from the National Research Foundation of Korea.


1. Pearl LH, Schierz AC, Ward SE, Al-Lazikani B, Pearl FM. Therapeutic opportunities within the PFI-3  DNA damage response. Nature reviews Cancer 2015;15:166-80
2. Holohan C, Van Schaeybroeck S, Longley DB, Johnston PG. Cancer drug resistance: an evolving paradigm. Nature reviews Cancer 2013;13:714-26
3. Powell SN, Bindra RS. Targeting the DNA damage response for cancer therapy. DNA Repair (Amst) 2009;8:1153-65
4. Lord CJ, Ashworth A. The DNA damage response and cancer therapy. Nature2012;481:287-94\5. Masliah-Planchon J, Bieche I, Guinebretiere JM, Bourdeaut F, Delattre O. SWI/SNF chromatin remodeling and human malignancies. Annu Rev Pathol 2015;10:145-71

6. Kadoch C, Crabtree GR. Mammalian SWI/SNF chromatin remodeling complexes and cancer: Mechanistic insights gained from human genomics. Sci Adv 2015;1:e1500447
7. Helming KC, Wang X, Roberts CWM. Vulnerabilities of mutant SWI/SNF complexes in cancer. Cancer Cell 2014;26:309-17
8. Ribeiro-Silva C, Vermeulen W, Lans H. SWI/SNF: Complex complexes in genome stability and cancer. DNA repair 2019;77:87-95
9. Park JH, Park EJ, Lee HS, Kim SJ, Hur SK, Imbalzano AN, et al. Mammalian SWI/SNF complexes facilitate DNA double-strand break repair by promoting
19gamma-H2AX induction. EMBO J 2006;25:3986-97

10. Watanabe R, Ui A, Kanno S, Ogiwara H, Nagase T, Kohno T, et al. SWI/SNF factors required for cellular resistance to DNA damage include ARID1A and ARID1B and show interdependent protein stability. Cancer research 2014;74:2465-75
11. Ogiwara H, Ui A, Otsuka A, Satoh H, Yokomi I, Nakajima S, et al. Histone acetylation by CBP and p300 at double-strand break sites facilitates SWI/SNF chromatin remodeling and the recruitment of non-homologous end joining factors. Oncogene 2011;30:2135-46
12. Shen J, Peng Y, Wei L, Zhang W, Yang L, Lan L, et al. ARID1A Deficiency Impairs the DNA Damage Checkpoint and Sensitizes Cells to PARP Inhibitors. Cancer Discov 2015;5:752-67
13. Velez-Cruz R, Manickavinayaham S, Biswas AK, Clary RW, Premkumar T, Cole F, et al. RB localizes to DNA double-strand breaks and promotes DNA end resection and homologous recombination through the recruitment of BRG1. Genes & development 2016;30:2500-12
14. Park JH, Park EJ, Hur SK, Kim S, Kwon J. Mammalian SWI/SNF chromatin remodeling complexes are required to prevent apoptosis after DNA damage. DNA Repair (Amst) 2009;8:29-39
15. Peng G, Yim EK, Dai H, Jackson AP, Burgt I, Pan MR, et al. BRIT1/MCPH1 links chromatin remodelling to DNA damage response. Nature cell biology 2009;11:865-72
16. Kakarougkas A, Ismail A, Chambers AL, Riballo E, Herbert AD, Kunzel J, et al.

Requirement for PBAF in transcriptional repression and repair at DNA breaks in actively transcribed regions of chromatin. Molecular cell 2014;55:723-32
17. Lee HS, Park JH, Kim SJ, Kwon SJ, Kwon J. A cooperative activation loop among20SWI/SNF, gamma-H2AX and H3 acetylation for DNA double-strand break repair. EMBO J 2010;29:1434-45

18. Kwon SJ, Park JH, Park EJ, Lee SA, Lee HS, Kang SW, et al. ATM-mediated phosphorylation of the chromatin remodeling enzyme BRG1 modulates DNA double- strand break repair. Oncogene 2014
19. Kwon SJ, Lee SK, Na J, Lee SA, Lee HS, Park JH, et al. Targeting BRG1 chromatin remodeler via its bromodomain for enhanced tumor cell radiosensitivity in vitro and in vivo. Molecular cancer therapeutics 2015;14:597-607
20. Fujisawa T, Filippakopoulos P. Functions of bromodomain-containing proteins and their roles in homeostasis and cancer. Nature reviews Molecular cell biology 2017;18:246-62
21. Zaware N, Zhou MM. Bromodomain biology and drug discovery. Nature structural & molecular biology 2019;26:870-9
22. Cochran AG, Conery AR, Sims RJ, 3rd. Bromodomains: a new target class for drug development. Nat Rev Drug Discov 2019;18:609-28
23. Filippakopoulos P, Picaud S, Mangos M, Keates T, Lambert JP, Barsyte-Lovejoy D, et al. Histone recognition and large-scale structural analysis of the human bromodomain family. Cell 2012;149:214-31
24. Filippakopoulos P, Knapp S. Targeting bromodomains: epigenetic readers of lysine acetylation. Nat Rev Drug Discov 2014;13:337-56
25. Filippakopoulos P, Qi J, Picaud S, Shen Y, Smith WB, Fedorov O, et al. Selective inhibition of BET bromodomains. Nature 2010;468:1067-73
26. Nicodeme E, Jeffrey KL, Schaefer U, Beinke S, Dewell S, Chung CW, et al.Suppression of inflammation by a synthetic histone mimic. Nature 2010;468:1119-2321

27. Moustakim M, Clark PGK, Hay DA, Dixon DJ, Brennan PE. Chemical probes and inhibitors of bromodomains outside the BET family. Medchemcomm 2016;7:2246-64

28. Perez-Salvia M, Esteller M. Bromodomain inhibitors and cancer therapy: From structures to applications. Epigenetics 2017;12:323-39

29. Clegg MA, Tomkinson NCO, Prinjha RK, Humphreys PG. Advancements in the Development of non-BET Bromodomain Chemical Probes. ChemMedChem 2019;14:362-85
30. Fedorov O, Castex J, Tallant C, Owen DR, Martin S, Aldeghi M, et al. Selective targeting of the BRG/PB1 bromodomains impairs embryonic and trophoblast stem cell maintenance. Sci Adv 2015;1:e1500723
31. Gerstenberger BS, Trzupek JD, Tallant C, Fedorov O, Filippakopoulos P, Brennan PE, et al. Identification of a Chemical Probe for Family VIII Bromodomains through Optimization of a Fragment Hit. Journal of medicinal chemistry 2016;59:4800-11
32. Lee SK, Park EJ, Lee HS, Lee YS, Kwon J. Genome-wide screen of human bromodomain-containing proteins identifies Cecr2 as a novel DNA damage response protein. Mol Cells 2012;34:85-91
33. Lee HS, Seo HR, Lee SA, Choi S, Kang D, Kwon J. BAP1 promotes stalled fork restart and cell survival via INO80 in response to replication stress. The Biochemical journal 2019;476:3053-66
34. Chou TC. Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacol Rev 2006;58:621- 81
35. Martin LJ, Koegl M, Bader G, Cockcroft XL, Fedorov O, Fiegen D, et al. Structure- Based Design of an in Vivo Active Selective BRD9 Inhibitor. Journal of medicinal22chemistry 2016;59:4462-75

36. Vangamudi B, Paul TA, Shah PK, Kost-Alimova M, Nottebaum L, Shi X, et al. The SMARCA2/4 ATPase Domain Surpasses the Bromodomain as a Drug Target in SWI/SNF-Mutant Cancers: Insights from cDNA Rescue and PFI-3 Inhibitor Studies. Cancer research 2015;75:3865-78
37. Papillon JPN, Nakajima K, Adair CD, Hempel J, Jouk AO, Karki RG, et al.

Discovery of Orally Active Inhibitors of Brahma Homolog (BRM)/SMARCA2 ATPase Activity for the Treatment of Brahma Related Gene 1 (BRG1)/SMARCA4- Mutant Cancers. Journal of medicinal chemistry 2018;61:10155-72
38. Bonner WM, Redon CE, Dickey JS, Nakamura AJ, Sedelnikova OA, Solier S, et al.GammaH2AX and cancer. Nat Rev Cancer 2008;8:957-67

39. Smith J, Tho LM, Xu N, Gillespie DA. The ATM-Chk2 and ATR-Chk1 pathways in DNA damage signaling and cancer. Adv Cancer Res 2010;108:73-112
40. Nigg EA. Mitotic kinases as regulators of cell division and its checkpoints. Nature reviews Molecular cell biology 2001;2:21-32
41. Lanz MC, Dibitetto D, Smolka MB. DNA damage kinase signaling: checkpoint and repair at 30 years. The EMBO journal 2019;38:e101801
42. Romero FA, Taylor AM, Crawford TD, Tsui V, Cote A, Magnuson S. Disrupting Acetyl-Lysine Recognition: Progress in the Development of Bromodomain Inhibitors. Journal of medicinal chemistry 2016;59:1271-98
43. Picaud S, Da Costa D, Thanasopoulou A, Filippakopoulos P, Fish PV, Philpott M, et al. PFI-1, a highly selective protein interaction inhibitor, targeting BET Bromodomains. Cancer research 2013;73:3336-4623