Inhibition of TXNRD or SOD1 overcomes NRF2-mediated resistance to β-lapachone

Alterations in the NRF2/KEAP1 pathway result in the constitutive activation of NRF2, leading to the aberrant induction of antioxidant and detoxification enzymes, including NQO1. The NQO1 bioactivatable agent β-lapachone can target cells with high NQO1 expression but relies in the generation of reactive oxygen species (ROS), which are actively scavenged in cells with NRF2/KEAP1 mutations. However, whether NRF2/KEAP1 mutations influence the response to β-lapachone treatment remains unknown. To address this question, we assessed the cytotoxicity of β-lapachone in a panel of NSCLC cell lines bearing either wild-type or mutant KEAP1. We found that, despite overexpression of NQO1, KEAP1 mutant cells were resistant to β-lapachone due to enhanced detoxification of ROS, which prevented DNA damage and cell death. To evaluate whether specific inhibition of the NRF2-regulated antioxidant enzymes could abrogate resistance to β-lapachone, we systematically inhibited the four major antioxidant cellular systems using genetic and/or pharmacologic approaches. We demonstrated that inhibition of the thioredoxin-dependent system or copper-zinc superoxide dismutase (SOD1) could abrogate NRF2-mediated resistance to β-lapachone, while depletion of catalase or glutathione was ineffective. Interestingly, inhibition of SOD1 selectively sensitized KEAP1 mutant cells to β-lapachone exposure. Our results suggest that NRF2/KEAP1 mutational status might serve as a predictive biomarker for response to NQO1-bioactivatable quinones in patients. Further, our results suggest SOD1 inhibition may have potential utility in combination with other ROS inducers in patients with KEAP1/NRF2 mutations. Highlights Aberrant activation of NRF2 in non-small cell lung cancer promotes resistance to β-lapachone via the antioxidant defense. Inhibition of the thioredoxin-dependent system and superoxide dismutase 1 increase sensitivity to β-lapachone treatment. Mutations in the NRF2/KEAP1 pathway might serve as predictive biomarker for response to β-lapachone in patients.


49
It is estimated that 38% of lung squamous cell carcinomas (SCC) and 18% of lung 50 adenocarcinomas (LuAD) harbor mutations in Nuclear factor erythroid 2-related factor 2 (NRF2), 51 or its negative regulator Kelch-like ECH-associated protein 1 (KEAP1) 1-3 , making this pathway 52 one of the most commonly mutated non-small cell lung cancer (NSCLC). The transcription 53 factor NRF2 acts as the primary cellular barrier against the deleterious effects of oxidative 54 stress by regulating the expression of cytoprotective genes. In healthy tissues, KEAP1 binds to 55 and harnesses the activity of NRF2, thereby promoting NRF2 ubiquitination and destruction by 56 the proteasome 4-6 . Loss-of-function mutations in KEAP1 and gain-of-function mutations in NRF2 57 found in NSCLC abolish this control and lead to constitutive NRF2 activity 1,7-9 . Cancer cells that 58 hijack NRF2 activity are equipped with a reinforced cytoprotective system through the induction 59 of antioxidant and drug detoxification pathways, thereby rendering them resistant to oxidative 60 stress and chemo/radio-therapy 10-12 .

62
High expression of the detoxification enzyme NAD(P)H:quinone oxidoreductase 1 (NQO1) is a 63 distinct biomarker of NRF2/KEAP1 mutant NSCLC tumors. NQO1 is a cytosolic flavoprotein that 64 catalyzes the two-electron reduction of quinones into hydroquinones in an effort to hamper 65 oxidative cycling of these compounds 13,14 . Although NQO1-dependent reduction of quinones 66 has been historically defined as a major detoxification mechanism, a number of quinones induce 67 toxicity following NQO1 reduction [15][16][17][18][19] . The mechanism behind this paradox relies on the 68 chemical properties of the hydroquinone forms. Unstable hydroquinones can be reoxidized to 69 the original quinone by molecular oxygen, which leads to the formation of superoxide radicals.

70
As the parent quinone is regenerated, the cycle continues, which amplifies the generation of 71 superoxide radicals, initiating a cascade of reactive oxygen species (ROS).

73
The ability of NQO1 to generate cytotoxic hydroquinones has been utilized as a strategy to 74 target cancer cells with high NQO1 levels. To date, β-lapachone and its derivatives are the most 75 studied NQO1-bioactivatable quinones, and the molecular mechanisms by which they promote 76 cytotoxicity have been thoroughly characterized 20-24 ( Figure 1A). NQO1 has been proposed as 77 a target for NSCLC therapy, as it is overexpressed in lung tumors but not in adjacent normal 78 tissues 25,26 . Thus, systemic delivery of β-lapachone would spare healthy lung tissue while 79 inducing robust cytotoxicity in tumor cells. Remarkably, a large fraction of NSCLC with high 80 NQO1 also harbor sustained NRF2 activation, which in turn could hinder the cytotoxic effects of 81 β-lapachone through the active scavenging of ROS. Therefore, although high levels of NQO1 could be exploited with a therapeutic intent in NRF2/KEAP1 mutant cancer cells, it is unclear 83 whether actions of NRF2 could limit β-lapachone efficacy. In this study, we aim to clarify 84 whether NQO1 represents a druggable strategy for NRF2/KEAP1 mutant NSCLC or, 85 conversely, these alterations promote resistance to β-lapachone.

88
Aberrant activation of NRF2 in NSCLC promotes resistance to B-lapachone

89
We compared the mRNA levels of NQO1 in healthy lung tissue, adenocarcinoma (LuAD) and 90 squamous cell carcinoma (LuSC) patients using the TGCA dataset ( Figure 1B)

103
In line with the mRNA data of LuSC and LuAD patients, KEAP1 MUT cell lines displayed uniformly 104 high NQO1 protein levels, while protein levels of NQO1 in KEAP1 WT cells were highly variable.

105
To determine the range of doses of β-lapachone that promote cell death in a NQO1-dependent 106 manner, cells were treated with increasing concentrations of β-lapachone alone or in 107 combination with the NQO1 inhibitor dicoumarol 30,31 ( Figure 1D). Additionally, we included in 108 the study Calu-3 cells, which harbor a polymorphic variant of NQO1 (NQO1*3) that results in 109 95% lower enzyme levels 32-34 ( Figure S1B). To recapitulate β-lapachone in vivo half-life 110 conditions 35 , cells were treated with β-lapachone for two hours and cell viability was analyzed 111 forty-eight hours after treatment. Our results showed that doses ranging from 1-6 µM induced 112 cell death in a dose-dependent and NQO1-specific manner. Remarkably, KEAP1 mutation 113 conferred resistance to β-lapachone treatment ( Figure 1D).

115
To test the ability of KEAP1 WT and KEAP1 MUT cell lines to promote redox cycling of β-lapachone, 116 we monitored the oxygen consumption rate (OCR) using the seahorse bioanalyzer. Basal OCR 117 was monitored prior injection of β-lapachone, and OCR was followed for 2 hours after β-118 lapachone addition. To validate whether changes in the OCR were a consequence of NQO1-119 dependent redox cycling of β-lapachone, we also monitored the OCR of cells co-treated with β-120 lapachone and dicoumarol. Treatment with β-lapachone resulted in a significant increase of the 121 OCR in both KEAP1 WT and KEAP1 MUT cells, which was precluded by the addition of dicoumarol 122 ( Figure 1E)

175
To sensitize KEAP1 MUT lung cancer cells to β-lapachone treatment, we sought to identify and 176 inhibit key NRF2-regulated antioxidant pathways. Given the major role of hydrogen peroxide in 177 mediating β-lapachone toxicity, we tested whether inhibition of individual peroxide detoxification 178 systems could overcome β-lapachone resistance ( Figure S3A). First, we tested the relevance 179 of catalase in the sensitivity to β-lapachone by using shRNAs (Figure 3A, 3B). We observed 180 that depletion of catalase did not affect the β-lapachone sensitivity of NSCLC cells, regardless 181 of KEAP1 mutational status. Of note, we found that H460 cells did not express detectable 182 catalase protein (Figure 3B, S3B).

184
NRF2 is a major upstream transcriptional regulator of enzymes involved in the thioredoxin-and

219
To validate these findings, we sough to test the effect of pharmacological inhibition of SOD1 on 220 β-lapachone efficacy. However, direct inhibitors of SOD1 that have been shown efficacy in cell-

226
First, we validated whether we could achieve SOD1 inhibition in cell culture using the copper 227 chelator ATN-224 ( Figure S4D). We observed that most of SOD1 activity was inhibited after 228 twenty-four-hour treatment using 2.5-5 µM. As expected, copper chelation did not affect SOD2

245
Aberrant NRF2 activation promotes resistance to therapeutics that rely on the production of 246 ROS, including multiple chemotherapeutics and radiation therapy. In this study, we find that 247 NRF2 activation also promotes resistance to the NQO1-activatable prodrug β-lapachone, which 248 relies on the generation of superoxide for its efficacy. While direct NRF2 inhibition could 249 potentially reverse this resistance, NRF2 inhibitors identified to date either lack specificity or potency. Further, the effects of NRF2 whole-body inhibition as anti-tumor strategy remain 251 unclear, as NRF2 activity in necessary for normal functioning of immune cells 45,46 .

253
Consequently, we have evaluated whether inhibition of the cellular antioxidant systems can 254 reverse the resistance of NRF2 active cells to ROS. We find that inhibitors of the TXN-255 dependent peroxide detoxification system and SOD1, but not glutathione or catalase depletion,

256
can reverse the resistance of KEAP1 MUT cells to β-lapachone. Surprisingly, we find that 257 KEAP1 MUT cells were highly resistant to auranofin compared to KEAP1 WT cells, which raises a 258 concern about the toxicity of the required auranofin doses to healthy tissues. The resistance of

505
Cell lysates were transferred to a microcentrifuge tubes, incubated at 90°C for 5 minutes, 506 followed by sonication to shred the DNA in a water bath sonicator (Diagenode). Samples were 507 centrifuged at 13,000 x rpm for 15 minutes at 4°C to precipitate the insoluble fraction. The 508 supernatant was transferred to a clean Eppendorf tube. Alternatively, total/gamma-H2A.X levels 509 were monitored in nuclear extracts (see protocol below). Cell lysates were mixed with 6X 510 sample buffer containing β-ME and separated by SDS-PAGE using NuPAGE 4-12% Bis-Tris 511 gels (Invitrogen), followed by transfer to 0.45µm Nitrocellulose membranes (GE Healthcare).

512
The membranes were blocked in 5% non-fat milk in TBS-T, followed by immunoblotting.

515
The following day, cells were washed with ice-cold PBS, collected in 1 ml of ice-cold PBS, 516 transferred to microcentrifuge tubes, and subjected to centrifugation at 13,000 × rpm for 1 min      Statistical analyses. Data were analyzed using a two-sided unpaired Student's t test.