Napabucasin (2-acetylfuro-1,4-naphthoquinone or BBI-608) is a small molecule currently being clinically evaluated in various cancer types. It has mostly been recognized for its ability to inhibit STAT3 signaling. However, based on its chemical structure, we hypothesized that napabucasin is a substrate for intracellular oxidoreductases and therefore may exert its anticancer effect through redox cycling, resulting in reactive oxygen species (ROS) production and cell death.
Binding of napabucasin to NAD(P)H:quinone oxidoreductase-1 (NQO1), and other oxidoreductases, was measured. Pancreatic cancer cell lines were treated with napabucasin, and cell survival, ROS generation, DNA damage, transcriptomic changes, and alterations in STAT3 activation were assayed in vitro and in vivo. Genetic knockout or pharmacologic inhibition with dicoumarol was used to evaluate the dependency on NQO1.
Napabucasin was found to bind with high affinity to NQO1 and to a lesser degree to cytochrome P450 oxidoreductase (POR). Treatment resulted in marked induction of ROS and DNA damage with an NQO1- and ROS-dependent decrease in STAT3 phosphorylation. Differential cytotoxic effects were observed, where NQO1-expressing cells generating cytotoxic levels of ROS at low napabucasin concentrations were more sensitive. Cells with low or no baseline NQO1 expression also produced ROS in response to napabucasin, albeit to a lesser extent, through the one-electron reductase POR.
Napabucasin is bioactivated by NQO1, and to a lesser degree by POR, resulting in futile redox cycling and ROS generation. The increased ROS levels result in DNA damage and multiple intracellular changes, one of which is a reduction in STAT3 phosphorylation.
Napabucasin is an orally administered small molecule currently undergoing clinical evaluation for treatment of cancer. It has been proposed to exert its anticancer activity by inhibiting STAT3 signaling and cancer stemness properties. Here, we show that napabucasin is a quinone that is bioactivated by oxidoreductases, in particular NAD(P)H:quinone oxidoreductase 1 (NQO1) and to a lesser extent the cytochrome P450 oxidoreductase (POR). Bioactivation of napabucasin generates cytotoxic levels of reactive oxygen species (ROS) resulting in DNA damage–induced cell death and multiple ROS-induced intracellular events, including a reduction in STAT3 phosphorylation. This better understanding of the mechanism of action of napabucasin will assist the development of novel, more effective therapeutic combination approaches, and will also aid in the identification of potential biomarkers of patients likely to respond to napabucasin.
Under physiologic conditions, incomplete reduction of oxygen results in the production of reactive oxygen species (ROS), including hydrogen peroxide (H2O2), the superoxide anion (O2−), and the hydroxyl radical (•OH). To protect molecules from ROS-induced damage, cells orchestrate a complex network of antioxidants to maintain proper cellular function. This reduction-oxidation (redox) balance is tightly controlled by several key transcription factors including nuclear factor erythroid-derived 2–like 2, NFE2L2/NRF2, which regulates the transcription of a number of target genes encoding components of antioxidant systems, glutathione synthesis enzymes, proteasome subunits, and heat-shock proteins (1–3). Disruption of this delicate redox balance has long been known to be associated with multiple diseases, including cancer development and progression (4). Tumors are thought to harbor a unique state of redox-regulatory mechanisms to support their pathologic survival and proliferation, demonstrating a biphasic response. At low levels, ROS are mutagenic and can promote tumor development by activating signaling pathways that regulate cellular survival, proliferation, differentiation, and metabolic adaptation. However, at high levels, ROS become toxic leading to oxidative stress and cell death or senescence (1, 5). To compensate for higher levels of intrinsic ROS, cancer cells have evolved adaptive mechanisms that increase their antioxidant capacity. NRF2 upregulation has been observed in multiple tumor types, and its expression has been shown to be required for pancreatic and lung cancer development (5–7). Thus, compared with normal cells, cancer cells with increased oxidative stress are likely more vulnerable to damage by further ROS insults, making modulation of tumor redox homeostasis an attractive therapeutic strategy.
The NRF2 target gene NAD(P)H:quinone oxidoreductase 1 (NQO1) is a two-electron oxidoreductase involved in the detoxification of quinones using NADH or NADPH to generate the corresponding hydroquinone derivative (8). Increased expression of NQO1 has been observed in many solid tumors, has been shown to occur early in tumorigenesis, and has been linked to multiple carcinogenic processes (9–15). For example, increased NQO1 expression is observed in precursor lesions (pancreatic intraepithelial neoplasia) and further increased expression occurs in invasive pancreatic ductal adenocarcinoma (13–15). The ability of NQO1 to generate hydroquinones, combined with its overexpression in many cancers, has been utilized as a therapeutic strategy, and various anticancer compounds that are bioactivated by NQO1 have been developed. Hydroquinones can exhibit toxicity through a number of mechanisms, depending on their chemical reactivity. Bioactivation of antitumor quinones such as mitomycin C or streptonigrin results in hydroquinone-mediated alkylation of DNA with interstrand crosslinking (16). In contrast, oxidoreduction of naphthoquinones, such as β-lapachone, results in an unstable hydroquinone that spontaneously reacts with oxygen to regenerate the original compound in a two-step back reaction, depleting NAD(P)H and generating substantial amounts of ROS (17, 18).
Napabucasin, also known as BBI-608, is an orally administered small molecule that is being clinically evaluated for the treatment of a variety of cancers, including pancreatic ductal adenocarcinoma (19, 20). It is mostly recognized for its ability to inhibit STAT3-mediated gene transcription with activity against bulk tumor cells and cancer stem cells, with inhibition of spherogenesis in vitro and tumor relapse in vivo (21–23). However, the mechanism by which napabucasin mediates these effects is not understood. In this report, we sought to further elucidate its mechanism of action based on the notion that napabucasin is a naphthoquinone (2-acetylfuro-1,4-naphthoquinone). We show that napabucasin is a substrate for NQO1, and to a lesser degree for the one-electron reductase cytochrome P450 reductase (POR). Bioactivation of napabucasin results in ROS generation, inducing oxidative stress and DNA damage with multiple ROS-induced intracellular events including, but not limited to, a reduction in STAT3 phosphorylation.
Materials and Methods
Cell lines were obtained from the ATCC or JCBR (Suit2) or generated from established human organoids as previously described (24) and cultured in DMEM (10-013-CV, Fisher Scientific) or RPMI (10-040-CV, Fisher Scientific) containing 10% FBS. All cells were cultured for no more than 20 passages and tested negative for mycoplasma using the MycoAlert Mycoplasma Detection Kit (LT07-318, Lonza). Cell line authentication was not performed.
NQO1 knockout CRISPR clones from MiaPaCa2, AsPc1 and DU145 cell lines were generated as previously described using Lenti_sgRNA_EFS_GFP (LRG) plasmids (Addgene #65656; refs. 25, 26). sgRNAs targeting unique locations at the NQO1 locus were designed, cloned, and validated by Sanger sequencing. Non-targeting sgRosa was used as a control. Cas9-expressing cells were infected and sorted for GFP expression on the FACSAria cell sorter (BD). For NQO1 knockout in FaDu cells, the parental cell line was transfected with ribonucleoprotein (RNP) complexes composed of sgRNA and Cas9NLS protein using the manufacturer's instructions (Thermo Fisher Scientific). In brief, functional sgRNA was generated by annealing tracrRNA and crRNA. A 1:1 ratio of sgRNA and Cas9NLS protein was mixed with LipoCas9 plus reagent and incubated for 5 minutes at room temperature to produce an RNP complex. The RNP complex was then mixed with Lipofectamine CRISPRMAX transfection reagent and added to the parental cell cultures. Following overnight incubation, the culture medium was replenished, and cells were expanded until a sufficient quantity of genomic DNA could be extracted. Successful gene editing was verified by heteroduplex analysis. Potential NQO1 knockout clones were selected, and complete NQO1 knockout was verified by Western blot. For expression of NQO1 in Panc1 cells, NQO1 was introduced by transfection of cDNA (Origene, RC200620) using XtremeGENE 9 (Roche, 06365787001) according to manufacturer's instructions. Functional assays were performed 36 hours after transfection with a cytomegalovirus CMV-driven GFP-expressing plasmid as control. In MDA-MB-231 cells, NQO1 was introduced using lentiviral transduction followed by blasticidin selection as directed by the manufacturer (GenTarget).
Expression and purification of NQO1
The coding sequence for human NQO1 was synthesized and cloned into pET15b (Novagen) using BamHI and NdeI restriction sites (Genewiz), along with an N-terminal hexahistidine affinity tag and thrombin cleavage site (MGSSHHHHHHSSGLVPRGSH). BL21(DE3)pLysS Escherichia coli (Promega) were transformed with plasmid and grown at 37°C in Luria–Bertani medium supplemented with 100 μg/mL ampicillin to an optical density at 600 nm of 0.8. Cultures were then chilled to 18°C, and protein expression was induced overnight with 0.5 mmol/L isopropyl β-D-1-thiogalactopyranoside. Cells were harvested, and lysate was loaded onto Ni-NTA affinity resin equilibrated in 50 mmol/L HEPES (pH 7) supplemented with 0.15 mol/L sodium chloride. Resin was washed extensively, and protein was eluted with buffer plus 0.25 mol/L imidazole. NQO1 was further purified with a Hiload 16/600 Superdex200 pg column (GE Healthcare); protein purity was judged to be >95% by SDS-PAGE. NQO1 was flash frozen for subsequent analysis.
Initial rates of NQO1 substrate digestion (0.4–25 μmol/L) were monitored using an assay in which the oxidation of NADPH to NADP+was quantified at 340 nm at 30°C using Spectramax 5 (Molecular Devices). Reactions of 0.02 μmol/L NQO1, 800 μmol/L NADPH in 50 mmol/L potassium phosphate (pH 7.4), and 5% DMSO with or without 5 mmol/L dicoumarol were initiated by addition of NADPH. Wells were monitored every 3 seconds for 2 minutes to obtain an initial linear signal that was converted to “μmol/L NADPH per minute per μmol/L NQO1” using a standard curve. Michaelis–Menten curves were generated with GraphPad Prism 5. Reactions were performed in triplicate. Similar reactions were carried out with purified NADPH:POR (C8113, Sigma), carbonyl reductase 1 (CBR1; ab85336, Abcam), and thioredoxin 1 (TRX1; ab51064, Abcam).
Napabucasin dose–response curves
Cells were plated at approximately 70% confluency, and increasing concentrations of napabucasin (range, 0.01–5 μmol/L) as single agent or combined with the antioxidant N-acetylcysteine (NAC), the ROS scavenger EUK-134 (Sigma), or the NQO1 inhibitor dicoumarol (Selleckchem) were added in triplicate 24 hours after plating and normalized to DMSO. Cell viability was assessed following 6 hours of treatment using CellTiter-Glo (Promega). Dose–response curves were generated using GraphPad Prism 5.
Measurement of ROS generation
ROS generation with simultaneous assessment of cell viability or changes in total to oxidized glutathione ratios following napabucasin treatment were determined by the ROS-Glo H2O2 (Promega) or GSH/GSSG-Glo assay (Promega), respectively, as per the manufacturer's instructions. In brief, cells were seeded in 96-well plates the day prior to treatment such that drug treatment was added when cells were 50% to 80% confluent. For ROS-Glo H2O2 assays, culture medium was replaced with 100 μL medium containing 25 μmol/L H2O2 substrate plus the desired drug concentration. After incubating for 6 hours at 37°C, 50 μL of supernatant were transferred to a new 96-well plate containing an equal volume of ROS detection reagent. A total of 50 μL CellTiter-Glo reagent (Promega) was added to the 96-well plate containing the remaining 50 μL of culture. For GSH/GSSG-Glo assays, cells were treated for 6 or 24 hours. Following treatment, medium was removed, and cells were washed with Hank's Balanced Salts and lysed with either total or oxidized glutathione reagent. Cell lysis was followed by luciferin generation and detection after which luminescence was read.
Measurements of glutathione (GSH) and GSH disulfide (GSSG) in snap-frozen tumor samples were done by adapting the procedures described by Moore and colleagues (27). Briefly, snap-frozen tissue specimens were lysed, incubated for 45 minutes at room temperature to allow derivatization of GSH to GSH-NEM after which supernatant was collected. For GSH measurements, 5 μL of derivatized sample was mixed with 50 μL of GSH-NEM standard ([13C2,15N]-glutathione, 200 μmol/L), vortexed, and transferred into autosampler glass vials. Similarly, sample extracts were added to an equal volume of GSSG internal standard solution ([13C4,15N2]-glutathione disulfide) for GSSG measurements. Samples were randomized in order to avoid bias due to machine drift and processed blindly. LC-MS analysis was performed using a Q Exactive HF mass spectrometer coupled to a Vanquish Horizon UHPLC system (Thermo Fisher Scientific). The acquired spectra were analyzed using XCalibur Qual Browser and XCalibur Quan Browser software (Thermo Fisher Scientific). Absolute quantification was performed by interpolation of the corresponding standard curve obtained from serial dilutions of commercially available standards run with the same batch of samples.
For ROS measurement by chloromethyl H2DCFDA, cells were washed with PBS, labeled with 5 μmol/L CM-H2DCFDA (ThermoFisher) for 30 minutes, and analyzed by flow cytometry.
In vivo subcutaneous transplantation
Nude mice were purchased from Charles River Laboratory (stock number 24102242), and 20 μL of 5.0 × 105MiaPaCa2 Rosa26 or MiaPaCa2 NQO1-71 cells mixed within an equal volume of PBS and Matrigel were injected subcutaneously. Tumor-bearing mice with a tumor volume of 150 mm3(0.5 × length × width2) were enrolled on a randomized basis to start treatment with either napabucasin dissolved in 0.5% methylcellulose or 0.5% methylcellulose. Mice were dosed once daily by oral gavage at 200 mg/kg for 24 days with monitoring of tumor volume every 3 days. All animal procedures were conducted in accordance with the Institutional Animal Care and Use Committee at Cold Spring Harbor Laboratory.
Western blot analysis
Whole-cell lysates were prepared at baseline or following 2 hours of drug treatment in a lysis solution of 20 mmol/L HEPES, 300 mmol/L NaCl, 5 mmol/L EDTA, 10% Glycerol, and 20% Triton X-100, pH 7.5, supplemented with Mini-complete protease inhibitors (11836170001, Roche) and a phosphatase inhibitor cocktail (4906845001, Roche). Standard procedures were followed for Western blotting using the following primary antibodies: Actin (8456, Cell Signaling Technology), STAT3 (9139, Cell Signaling Technology), pSTAT3 (9145, Cell Signaling Technology), pJAK1 (3331, Cell Signaling Technology), JAK1 (MAB42601-SP, R&D), pJAK2 (3771, Cell Signaling Technology), JAK2 (3230, Cell Signaling Technology), NQO1 (3187, Cell Signaling Technology), POR (ab13513, Abcam), β-Tubulin (2148, Cell Signaling Technology), Catalase (12980, Cell Signaling Technology), and NRF2 (ab62352, Abcam). Proteins were detected using horseradish peroxidase–conjugated secondary antibodies (Jackson ImmunoResearch Laboratories).
Cells were fixed with 3.7% formaldehyde, permeabilized with 0.1% Triton-X100, blocked with 0.1% BSA, and incubated for 1 hour at room temperature with phospho-histone H2A.X antibody (9718, Cell Signaling Technology) followed by Alexa488 or Alexa647-labeled secondary antibody and DAPI as counterstain. Imaging was performed with a Leica TCS SP8 laser scanning confocal microscope (Boulder Grove Il).
RNA-sequencing and analysis
Following 2 hours of treatment with 0.5 μmol/L napabucasin or DMSO, cells were lysed using TRIzol Reagent (15596-018; Thermo Fisher Scientific), and RNA was extracted with a PureLink RNA mini kit (12183018A; Thermo Fisher Scientific). Libraries were prepared using a KAPA mRNA HyperPrep Kit for Illumina sequencing (Roche, KR1352–v4.17) according to the manufacturer's instructions, and single-end RNA-sequencing was performed on an Illumina NextSeq500. All RNA-sequencing data are available at the Gene Expression Omnibus under the accession number GSE135352.
Differential gene expression analysis was performed using Bioconductor package DESeq2 (28), with a prefiltering step to remove genes that had no reads or reads only in one sample. Only genes with an adjusted P value < 0.05 and a log2 fold change ≥ 1 were retained as significantly differentially expressed. Gene set enrichment analysis (29) was performed to evaluate napabucasin-mediated alterations in the HALLMARK IL6-JAK-STAT3 geneset specifically. Additional functional enrichment analysis was performed by creating protein–protein and Reactome pathway–protein interaction networks using Search Tool for Retrieval of Interacting Genes/Proteins (STRING) version 11.0 (30), stringApp version 1.4.2 (31), and Cytoscape version 3.7.1 (32).
Synthetic, siRNA oligos targeting NQO1, NQO2, POR, Ferredoxin Reductase (FDXR), Cytochrome B5 Reductase 1 (CYB5R1), Cytochrome B5 Reductase 3 (CYB5R3), Cytochrome B5 Reductase 4 (CYB5R4), CBR1, and Thioredoxin Reductase 1 (TXNRD1) were obtained from Ambion. Cells were transfected with siRNA using Lipofectamine RNAiMAX (Invitrogen) and assayed at 72 hours after transfection.
Samples were lysed with TRIzol Reagent, with homogenization for snap-frozen tumor samples, and RNA was extracted with a PureLink RNA mini kit (12183018A; Thermo Fisher Scientific) followed by reverse transcription of 1 μg RNA using TaqMan reverse transcription reagents (N808-0234; Applied Biosystems). qPCR was performed using gene-specific TaqMan probes (Applied Biosystems) and master mix (4440040; Applied Biosystems). Gene expression was normalized to HPRT or ACTIN. siRNA knockdown was verified by qPCR with RT-qPCR primers (Qiagen) and iTaq universal SYBR green supermix (Bio-Rad) on a CFX connect real-time system (Bio-Rad).
Napabucasin activity and ROS generation
Given that napabucasin was originally hypothesized to target cancer cells and cancer stem cells by reduction of STAT3 signaling (20, 21), we first determined whether these activities were also observed in a panel of pancreatic cancer cell lines. When cells were treated for 6 hours with increasing concentrations of napabucasin, differential cytotoxicity was observed (Fig. 1A). Reductions in the active, phosphorylated form of STAT3, as well as phosphorylated JAK1 and JAK2, were observed, but to a different degree for each cell line (Fig. 1B). Based on its naphthoquinone structure (Supplementary Fig. S1A), we hypothesized that napabucasin may function as a ROS generator through NQO1-mediated redox cycling, and that reduced JAK/STAT signaling may be a downstream effect of napabucasin-mediated ROS production. Indeed, treatment with napabucasin increased ROS levels and reduced cell viability (Fig. 1C–F), which was mitigated by the addition of the antioxidant NAC (Fig. 1G). Of note, cells in which relative levels of napabucasin-induced ROS were higher (MiaPaCa2 and AsPc1) were found to be more sensitive to napabucasin compared with those with less napabucasin-induced ROS generation (Suit2 and Panc1), although higher concentrations of napabucasin were required for AsPc1 cells (Fig. 1D and E). Nevertheless, generation of ROS, as measured by a change in the ratio of the antioxidant GSH to its oxidized species (GSSG), in response to a fixed, low, dose of napabucasin (0.5 μmol/L), correlated with response (Fig. 1F). Similar observations were made in colon and lung cancer cells (Supplementary Fig. S1B), with a rescue in cell viability when napabucasin was combined with the ROS scavenger EUK-134 (Supplementary Fig. S1C).
Napabucasin is an NQO1 substrate
To determine whether napabucasin can act as a substrate for NQO1-mediated reduction using NADPH, we assessed NQO1 substrate digestion in a cell-free system in which we quantified the oxidation of NADPH to NADP+when NQO1 was incubated with either napabucasin or the known NQO1 substrate β-lapachone as a control (18, 33). Napabucasin was shown to directly bind to human NQO1 with high catalytic activity. This effect was blocked by dicoumarol, a specific NQO1 inhibitor that competes with NADH/NADPH substrate binding (Fig. 2A). Moreover, compared with β-lapachone, napabucasin had tighter NQO1-binding affinity (KM) and better catalytic efficiency (kcat/KM; Fig. 2A), suggesting that napabucasin is a more potent substrate of NQO1.
We next evaluated NQO1 expression in a panel of pancreatic cancer cell lines (Fig. 2B) and whether pharmacologic inhibition of NQO1 could reverse the napabucasin-mediated effects. Combined treatment of napabucasin and dicoumarol rescued cell viability in MiaPaCa2, AsPc1, and organoid-derived pancreatic cancer cell lines (Fig. 2C, Supplementary Fig. S2A). In contrast, combination treatment did not rescue viability in Suit2 or Panc1 cells (Fig. 2C), due to the undetectable levels of NQO1 protein in these cells (Fig. 2B). To determine whether dicoumarol also prevented napabucasin-mediated ROS production, we measured H2O2 levels and the GSH:GSSG ratio in cells treated with napabucasin and/or dicoumarol. The dicoumarol-mediated rescue in cell viability inversely correlated with changes in the levels of ROS generation: napabucasin-mediated increases in ROS levels were inhibited by dicoumarol in MiaPaCa2, AsPc1, and organoid-derived cell lines, but not in the NQO1-deficient Suit2 or Panc1 cell lines (Fig. 2D and E; Supplementary Fig. S2B).
To further assess the dependency of napabucasin activity on NQO1, we used CRISPR/Cas9 to knock out NQO1 in the pancreatic cancer cell lines MiaPaCa2 and AsPc1 (Fig. 3; Supplementary Fig. S3A–S3C), as well as DU145 cells, a metastatic prostate cancer cell line (Supplementary Fig. S3D and S3E), and FaDu cells, a hypopharynx squamous cell carcinoma cell line (Supplementary Fig. S3D and S3F). NQO1 ablation made cells more resistant to napabucasin (2.5–3.5-fold) and to a lesser degree to β-lapachone (1.5-fold; Fig. 3A and B; and Supplementary Fig. S3A). The reduced activity of napabucasin in NQO1-ablated cells was associated with decreased ROS induction, as measured by H2O2 levels (Fig. 3C) and a shift in the GSH:GSSG ratio (Fig. 3D). Similar observations were made in subcutaneous xenografts, with intratumoral ROS generation, as detected by reduced GSH:GSSG ratios, in MiaPaCa2 Rosa26 xenografts treated with napabucasin but not in tumors derived from NQO1 knockout cells (Supplementary Fig. S3B and S3C). The NQO1 dependency of napabucasin was also observed in DU145 and FaDu cells (Supplementary Fig. S3D–S3F). Similarly, ectopic expression of NQO1 in the NQO1-negative Panc1 cells and an NQO1-negative breast cancer cell line MDA-MB-231 sensitized these cells to napabucasin with an associated increase in ROS production (Fig. 3E and F; and Supplementary Fig. S3G).
Taken together, these results show that napabucasin induces ROS in human tumor cells in an NQO1-dependent manner, and suggest that napabucasin-mediated cytotoxicity may be dependent on both ROS and NQO1 expression.
Napabucasin activity and changes in STAT3 signaling
Given the role of NQO1 and ROS in napabucasin-mediated cytotoxicity, and the observed decrease in phosphorylation of STAT3 (pSTAT3) upon napabucasin treatment (Fig. 1B), we sought to determine whether NQO1 expression and ROS generation were required to inhibit activation of the STAT3 pathway. In MiaPaCa2 cells, with high baseline pSTAT3, we found that NQO1 knockout predominantly restored pSTAT3 expression in napabucasin-treated cells, as did the addition of the NQO1 inhibitor dicoumarol (Fig. 3G). However, in AsPc1 cells that have much lower basal levels of pSTAT3 (Fig. 1B), napabucasin treatment did not diminish pSTAT3 levels in an NQO1-dependent manner (Supplementary Fig. S3H). Similarly, while there were no changes in pSTAT3 upon treatment of the NQO1-deficient MDA-MB-231 breast cancer cells with napabucasin, the reintroduction of NQO1 to MDA-MB-231 cells was sufficient to restore the ability of napabucasin to diminish pSTAT3 levels (Supplementary Fig. S3G). In addition, treatment with H2O2 was sufficient to reduce pSTAT3 expression in all pancreatic cancer cell lines (Fig. 3H; Supplementary Fig. S3I), and pSTAT3 levels were partially restored in MiaPaCa2 cells when napabucasin was combined with NAC (Fig. 3H). These data indicate that napabucasin-mediated inhibition of STAT3 activity is a secondary effect from the treatment-induced high levels of ROS, which is, in part, dependent on NQO1 expression.
Napabucasin-induced transcriptomic changes
Based on the notion that napabucasin induces ROS in an NQO1-dependent manner, resulting in ROS-driven intracellular signaling modifications, we further evaluated the transcriptomic changes following 2 hours of treatment with napabucasin in MiaPaCa2 cells, two NQO1 knockout clonal lines (NQO1-71 and NQO1-163), and the respective Rosa26 control. In the parental MiaPaCa2 cells, a total of 158 genes were differentially expressed, with the majority of genes being upregulated following treatment with napabucasin (Supplementary Fig. S4A; Supplementary Table S1). Of those 158 genes, 24 showed an NQO1-dependent differential expression, including many genes known to be induced upon cellular stress (Fig. 4A). Surprisingly, there was no significant, NQO1-dependent enrichment of the JAK-STAT signaling pathway, with only 3 genes from the JAK-STAT geneset significantly enriched in the napabucasin-treated parental MiaPaCa2 cells (HMOX1, MAP3K8, SOCS3; FDR corrected P = 0.02; Supplementary Fig. S4B). Heme oxygenase (HMOX1) is a well-known NRF2 target gene, and its expression is known to be induced by ROS to protect cells against oxidative damage by catalyzing the breakdown of heme molecules and sequestering the redox-active Fe2+(3, 34, 35). HMOX1 expression was strongly induced upon treatment with napabucasin in an NQO1-dependent manner, both in vitro and in vivo, with increased expression of HMOX1, as well as other NRF2 target genes, in the NQO1-positive MiaPaCa2 and AsPC1 cells (Fig. 4A–C) or tumors from MiaPaCa2 xenografts (Fig. 4D), but not in the NQO1 knockout MiaPaCa2 cells or xenografts or the NQO1-negative Suit2 and Panc1 cells (Fig. 4A, C, and D).
Additional protein–protein (Supplementary Fig. S4C) and pathway–protein interaction network (Fig. 4B) analysis with the differentially expressed genes in the parental MiaPaCa2 cells further highlighted the induction of oxidative stress and DNA damage upon treatment with napabucasin, with upregulation of the stress response genes ATF3 and ATF4, as well as other members of the AP1 transcription complex (FOS, JUN) and early response genes involved in cell-cycle arrest in response to DNA damage (CDKN1A, BTG1, BTG2; Fig. 4B). This ROS-induced stress response upon treatment with napabucasin was seen across cell lines and in the MiaPaCa2 Rosa26, but not in the MiaPaCa2 NQO1 knockout, xenografts (Fig. 4D and E).
Napabucasin and cytochrome P450 oxidoreductase
Based on the observation that napabucasin still has an effect in NQO1-deficient cells, with a reduction in cell viability and ROS generation, albeit at a lesser degree compared with NQO1-expressing cells, we hypothesized that the antitumor effects of napabucasin may also be conferred via NQO1-independent pathway(s). Indeed, there are several non-NQO1 reductases with the potential to generate ROS from quinones (Supplementary Fig. S5A; refs. 36, 37). To this end, we examined the interactions between napabucasin (and β-lapachone) and a number of one-electron reductases: POR, CBR1, and TRX1. In a cell-free system, both napabucasin and β-lapachone were shown to be substrates of POR (Fig. 5A). Additional evaluation showed that napabucasin and β-lapachone have different specificities for the other reductases studied. For example, although both napabucasin and β-lapachone can be efficiently reduced by NQO1 and POR, β-lapachone can also be reduced by CBR1, whereas CBR1 has little activity against napabucasin (Fig. 5B).
To determine whether POR can substitute for NQO1 as the reductase that acts on napabucasin in NQO1-deficient cells, we used siRNA to deplete various reductases in Panc1 cells, which do not express detectable NQO1 protein but do express POR (Fig. 2B; Supplementary Fig. S5B and S5C). Of the siRNAs screened, siRNA directed against POR inhibited napabucasin-mediated cell death to the greatest extent (Fig. 5C), with an associated reduction in ROS generation (Fig. 5D). Interestingly, knockdown of some oxidoreductases sensitized Panc1 cells to napabucasin, an effect most profoundly observed with knockdown of the NRF2 target gene TXNRD1 (Fig. 5C). The increased sensitivity to napabucasin seen when TXNRD1 was knocked down was accompanied by elevated ROS production (Fig. 5D). These results highlight the intricate regulation of intracellular oxidative stress and suggest that in the absence of NQO1, napabucasin may be a substrate for POR, which can generate ROS and mediate cell death. Conversely, other cellular reductases (e.g., TXNRD1) may function as antioxidants, inhibiting the cytotoxic activity of napabucasin.
In conclusion, our data indicate that napabucasin is bioactivated by NQO1, with a role for the one-electron reductase POR in cells that do not express NQO1. This, in turn, results in increased ROS generation causing DNA damage (Fig. 6A; Supplementary Fig. S6) and a multitude of intracellular events including a reduction in STAT3 phosphorylation, stabilization of NRF2 (Supplementary Fig. S7) with upregulation of NRF2 target genes as well as the activation of other stress-induced genes and protective mechanisms in an attempt to counteract the ROS-induced damage (Fig. 6B). Given the redox difference between cancer cells and normal cells, the high expression of NQO1 in many cancers, including pancreatic cancer, makes disruption of this balance by napabucasin an attractive, tumor-specific approach.
Here, we show that the naphthoquinone napabucasin can be bioactivated by the cellular reductases NQO1 and, to a lesser extent, POR, resulting in the production of ROS and disruption of the cellular redox balance, resulting in DNA damage–induced cell death. Although traditionally ROS are considered to be toxic molecules causing indiscriminate damage to proteins, nucleic acids, and lipids, it is increasingly recognized that they also play a significant role as secondary messengers in cellular signaling (37). A number of transcription factors contain redox-sensitive cysteine residues at their DNA binding sites, including NF-κB, HIF-1, and p53. In addition, ROS can either inhibit or activate protein function through altering their phosphorylation status via thiol oxidation of either tyrosine phosphatases or kinases (1, 38, 39). Similar to previous reports (21–23), we observed a decrease in STAT3 phosphorylation upon treatment with napabucasin in pancreatic and breast cancer cells. However, the ability to do so appeared to be dependent on NQO1 and ROS generation. Indeed, STAT3 phosphorylation can be inhibited directly or indirectly by ROS (40, 41), but in the absence of a reduction in JAK-STAT signaling in response to napabucasin, the functional importance of the reduction in pSTAT3 expression remains unclear. Instead, the decrease in pSTAT3 is most likely a secondary event in response to increased ROS and may serve as a pharmacodynamic biomarker in which high baseline pSTAT3 expression may also be predictive of response. Consistently, early data have shown improved survival in patients with advanced, pSTAT3-positive colorectal cancer treated with napabucasin compared with placebo (42).
Redox alterations in cancer cells are complex, and cancer cells have become adapted to higher levels of oxidative stress resulting in malignant transformation, metastasis, and drug resistance. Drug-resistant cancer cells may use redox-regulatory mechanisms to promote cell survival and tolerate external insults from anticancer agents. Therapeutically increasing ROS levels by agents such as napabucasin may cause cells to lose their “stemness,” rendering them drug sensitive (43). Although this currently is an unexplored area, it is evident that ROS generation plays a critical role in the antitumor activity of napabucasin. The use of NQO1 as a predictive biomarker for sensitivity to napabucasin, or other quinone anticancer drugs, is appealing. However, NQO1 protein levels are not stable and can for example be induced by a host of dietary components or environmental factors (16). In addition, we observed a differential response to napabucasin also within the NQO1-positive cells. In particular, AsPc1 cells required the highest drug concentration to induce ROS-mediated cell death with temporal changes in response when cells were treated for a longer period of time (data not shown). Despite of previous reports indicating expression of the antioxidant catalase as important mechanism of resistance to β-lapachone in NQO1-positive cells (33, 44), we did not observe such a correlation with regards the response to napabucasin (Supplementary Fig. S7A). Gene and protein expression analysis however showed marked NRF2 pathway activation after only 2 hours of drug exposure (Fig. 4; Supplementary Fig. S7B), and ROS-induced upregulation of various cytoprotective mechanisms may play a role in the temporal kinetics of response. Moreover, we observe that cells that do not express NQO1 are still able to generate ROS following napabucasin treatment, although to a lesser degree, through POR, an oxidoreductase known to be the source of ROS generation resulting in paraquat-induced cell death (36). The precisely coordinated and dynamic regulation of ROS generation and detoxification is further highlighted by the different effects of napabucasin when expression of various oxidoreductases is reduced by siRNA. In particular, reduction of the antioxidant TXNRD1 enhanced napabucasin activity and concomitantly increased ROS production. Thus, a more comprehensive “redox-signature” may be better predictive of tumors likely to respond to napabucasin, rather than the expression of a single protein.
The thioredoxin system is an important thiol antioxidant, consisting of thioredoxin (TRX) and thioredoxin reductase, frequently upregulated in cancer (45). To maximally exploit ROS-mediated cell death mechanisms, combining napabucasin with agents that inhibit the thioredoxin pathway, such as sulfasalazine or auranofin (46, 47), may further enhance its antitumor activity. Many conventional cytotoxic cancer drugs can also directly or indirectly increase ROS levels in cancer cells and may synergize with napabucasin. Current clinical trials are testing this hypothesis. For instance, the combination of napabucasin, gemcitabine, and nab-paclitaxel is currently being evaluated as a treatment for metastatic pancreatic cancer (CanStem111P, NCT02993731; ref. 19). However, as with all anticancer therapies, identifying responsive subgroups is paramount in order to significantly improve clinical outcomes. Our study provides important insights regarding the mechanism of action of napabucasin, which will assist further biomarker development and research aimed to identify optimal therapeutic combination approaches with identification of those patients who are most likely to benefit from napabucasin.
Disclosure of Potential Conflicts of Interest
J. Li is an employee/paid consultant for Boston Biomedical Inc. A.-Y. Chang is an employee/paid consultant for Boston Biomedical Inc. H.A. Rogoff is an employee/paid consultant for Boston Biomedical Inc. J.D. Watson is an unpaid consultant/advisory board member for Boston Biomedical Inc. D.A. Tuveson is an employee/paid consultant for and holds ownership interest (including patents) in Surface Oncology and Leap Therapeutics, and reports receiving commercial research grants from Fibrogen and ONO. No potential conflicts of interest were disclosed by the other authors.
Conception and design: F.E.M. Froeling, I.I.C. Chio, A.-Y. Chang, L.C. Trotman, H.A. Rogoff, J.D. Watson, D.A. Tuveson
Development of methodology: F.E.M. Froeling, M.M. Swamynathan, I.I.C. Chio, J. Li, P. Belleau, D.A. Tuveson
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F.E.M. Froeling, M.M. Swamynathan, I.I.C. Chio, E. Brosnan, M.A. Yao, P. Alagesan, M. Lucito, J. Li, A.-Y. Chang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F.E.M. Froeling, M.M. Swamynathan, A. Deschênes, I.I.C. Chio, L.C. Trotman, P. Belleau, H.A. Rogoff, D.A. Tuveson
Writing, review, and/or revision of the manuscript: F.E.M. Froeling, M.M. Swamynathan, A. Deschênes, I.I.C. Chio, A.-Y. Chang, Y. Park, H.A. Rogoff, J.D. Watson, D.A. Tuveson
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): I.I.C. Chio, P. Alagesan, A.-Y. Chang
Study supervision: Y. Park, H.A. Rogoff, J.D. Watson, D.A. Tuveson
We would like to thank the Cold Spring Harbor Cancer Center Support Grant (CCSG) shared resources: E. Ghiban in the Next Generation Sequencing Core Facility, P. Moody and C. Kanzler in the Flow Cytometry Facility, S. Costa at Mass Spectrometry Core Facility, Q. Gao at Histology Core Facility, and all staff at the Animal Facility. The CCSG is funded by the NIH Cancer Center Support Grant 5P30CA045508. This work was supported by the Lustgarten Foundation, where D.A. Tuveson is a distinguished scholar and Director of the Lustgarten Foundation–designated Laboratory of Pancreatic Cancer Research. D.A. Tuveson is also supported by the Cold Spring Harbor Laboratory Association and the David Rubinstein Center for Pancreatic Cancer Research at MSKCC, the V Foundation, the Thompson Foundation, and the Simons Foundation (552716). In addition, this work was supported by the NIH P30CA045508, P50CA101955, P20CA192996, U10CA180944, U01CA168409, U01CA210240, R33CA206949, R01CA188134, and R01CA190092 to D.A. Tuveson. We are also grateful for support from the Donaldson Charitable Trust for F.E.M. Froeling and The Northwell Health Affiliation for F.E.M Froeling and D.A. Tuveson. Y. Park is supported by the NIH (R50CA211506). I.I.C. Chio is supported by Pancreatic Cancer Action Network (PG009667 - PANCAN 18-35-CHIO); the V Foundation (PG009685 - VFND V2018-017); and Columbia University Medical Center (Paul Marks Scholar Award). We thank Dr. Lindsey Baker and Dr. Claudia Tonelli for critical review of the article. We also thank Dr. Tom Miller for the original observation that napabucasin is a quinone.
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.