Aromatase Inhibitor–Mediated Downregulation of INrf2 (Keap1) Leads to Increased Nrf2 and Resistance in Breast Cancer Free

Drug resistance is the major obstacle to the successful treatment of many cancers (1). The factors that contribute to the development of drug resistance include alterations in drug intake, efflux, metabolism, and excretion. Deregulation of cell death by evasion of apoptosis, necrosis, mitotic catastrophe, or senescence also contributes to drug resistance (1–3). In addition, the differential expression of membrane proteins such as solute carriers, channels, and ATP-binding cassette transporters have all been demonstrated to play important role in drug resistance (4, 5).

Breast cancer is the most common cancer among women (6). Aromatase inhibitors are an effective first line of treatment for ERα-positive breast cancer that constitutes three-fourth of all types of breast cancers (7). Aromatase (cytochrome P450 CYP19A1) catalyzes the rate-limiting and final step of estrogen biosynthesis; the aromatization of androgens to estrogens (8, 9). Breast cancer tissues have been shown to express aromatase and produce higher levels of estrogens than noncancerous cells (7). Estrogens stimulate breast cancer cell growth and proliferation. Aromatase inhibitors became the choice of treatment for breast cancer in postmenopausal women because they block the synthesis of estrogens required by cancer cells to grow (10). Currently, there are three aromatase inhibitors approved by the FDA, letrozole, anastrozole, and exemestane (Supplementary Fig. S1). These are approved for postmenopausal women with hormone receptor–positive breast cancer in both the adjuvant and metastatic setting. Letrozole is more potent than other aromatase inhibitors in reducing plasma estrogen levels (11). While aromatase inhibitors are a very effective treatment, their benefit is often limited by the emergence of resistance that occurs in a significant number of patients in the adjuvant setting and is inevitable in the metastatic setting.

The INrf2 (Keap1):Nrf2 complex acts as a cellular sensor of xenobiotics, drugs, and radiation-induced ROS/electrophilic stress (12). Nuclear factor Nrf2 controls the expression and coordinated induction of a battery of genes encoding detoxifying enzymes [quinone oxidoreductases (NQO1 and NQO2), glutathione S-transferases, heme oxygenase 1 (HO-1)], glutathione and related proteins [glutathione, thioredoxins, γ-glutamyl cysteinyl synthetase (γ-GCS)], ubiquitination enzymes and proteasomes (12, 13), drug transporters (MRP; refs. 14, 15), and antiapoptotic proteins (16). Nrf2 is retained in the cytoplasm by an inhibitor INrf2 or Keap1 (17, 18). INrf2 functions as an adapter for Cul3/Rbx1-mediated degradation of Nrf2 (12). In response to chemical/drug/radiation including antioxidant tert-butyl hydroquinone (t-BHQ)-induced oxidative/electrophilic stress, Nrf2 is switched on (separation from INrf2 and stabilization of Nrf2) and then off (ubiquitination and degradation of Nrf2) by distinct early and delayed mechanisms (12). Oxidative/electrophilic modifications of INrf2 cysteine151 and/or PKC phosphorylation of Nrf2 serine40 result in the escape or release of Nrf2 from INrf2 (12). Nrf2 is stabilized and translocates to the nucleus, forms heterodimers with small Maf or Jun proteins, and binds antioxidant response elements resulting in coordinated activation of gene expression (12). Indeed, in vivo evidence has demonstrated the importance of Nrf2 in protecting cells from the toxic and carcinogenic effects of many environmental insults. Nrf2-knockout mice were susceptible to acute damages induced by acetaminophen, ovalbumin, cigarette smoke, and pentachlorophenol and had increased tumor formation when exposed to carcinogens such as benzo[a]pyrene, diesel exhaust, and N-nitrosobutyl (4-hydroxybutyl) amine (19–22). Therefore, Nrf2 appears to play a significant role in cytoprotection and cell survival (12). In addition, Nrf2 plays significant role in prevention of cancer metastasis (23–25).

Studies have also described the detrimental effects of Nrf2 (26–30). Persistent stabilization and nuclear accumulation of Nrf2 is suggested to play a role in survival of cancer cells and drug resistance. Increase in Nrf2 due to inactivating mutations in INrf2 has been reported in lung cancer (26, 27). Although Nrf2 is thought to contribute to drug resistance by inducing cytoprotective proteins (28, 29), its role in resistance of breast cancer to aromatase inhibitors remains unknown.

The studies in this report showed that aromatase inhibitor–resistant breast cancer cells contain lower INrf2 and higher Nrf2 levels, as compared with drug-sensitive cells. Studies also revealed that higher Nrf2 was due to decreased INrf2, lower ubiquitination, and slower degradation of Nrf2 in aromatase inhibitor–resistant cells. Higher Nrf2-mediated increase in biotransformation enzymes, drug transporters, and antiapoptotic proteins contributed to reduced efficacy of drugs and prevention of apoptosis that led to drug resistance. Interestingly, LTLT Ca cells deficient in Nrf2 (LTLTCa-Nrf2KD) showed reduced levels of aldehyde dehydrogenase (ALDH), a marker of tumor-initiating cells (TIC), significantly decreased mammosphere formation and increased sensitivity to exemestane and doxorubicin, as compared with parental LTLTCa cells expressing higher levels of Nrf2. These results collectively suggest that persistent aromatase inhibitor treatment downregulated INrf2 leading to higher Nrf2 and downstream cytoprotective proteins that resulted in increased aromatase inhibitor drug resistance.

Puromycin dihydrochloride (sc-108071), control shRNA lentiviral particles-A (sc-108080), Nrf2 shRNA (sc-37030-V), anti-Nrf2 (sc-13032), anti-Keap1 (sc-15246), anti-HO-1 (sc-10789), anti-NQO1 (sc-32793), anti-Bcl-2 (sc-492), anti-Bcl-xL (sc-8392), anti-Mcl-1 (sc-819), anti-Lamin B (sc-6217), anti-Mdr-1 (sc-8318), anti-MRP1 (sc-13960), anti-HER2 (sc-284), anti-Ub (sc-8017), anti-Ku70 (sc-17789) antibodies were from Santa Cruz Biotechnology. Glutathione assay kit (item No. 703002) was from Cayman Chemical. Ultra-low attachment of 24-well plate (cat. no. 3473) for mammospheres was obtained from Corning. DCFDA Cellular ROS Detection Assay Kit (cat. no. ab113851) and γ-glutamylcysteine synthetase (GCLC, ab40929) antibody were obtained from Abcam. Anti-LDH (cat. no. 3558) from Cell Signaling Technology, anti-MRP4 (cat. no.ALX-801-038) from Enzo Life Sciences, anti-BCRP (cat. no. OP191-200UL), Ku80 (cat. no. NA54), and proteasome inhibitor MG-132 (cat. no. 474790) from Millipore were purchased for Western blotting. Aldefluor assay kit was obtained from Stem Cell Technologies. Aromatase inhibitors (letrozole and anastrozole) were provided by Dr. Brodie's laboratory.

Aromatase inhibitor–sensitive cells (MCF-7Ca and AC1) and aromatase inhibitor–resistant cells (LTLTCa and AnaR) have been described previously (31–33). Briefly, human breast cancer MCF-7 cells were stably transfected with the human aromatase gene to generate MCF-7Ca cells (32). Letrozole-resistant LTLTCa cells were isolated from MCF-7Ca mouse xenograft tumors treated with letrozole for 56 weeks. The lower expression of ERα in aromatase inhibitor–resistant cells (LTLTCa and AnaR) as compared with aromatase inhibitor–sensitive cells was previously described (34). However, we did observe as is already published (31) that LTLTCa cells express significantly less ERα than MCF-7Ca cells. Similar to MCF-7Ca and LTLTCa cells, AC1 cells were generated from the MCF-7 cells by stable transfection with human aromatase gene. AnaR cells were anastrozole-resistant cells isolated form AC1 mouse xenograft tumors treated with anastrozole for 14 weeks (31). MCF-7Ca cells were grown in DMEM containing 700 μg/mL G418 sulfate and 5% FBS. DMEM with 700 μg/mL G418 sulfate and 10% FBS were used to culture AC1 cells. LTLTCa cells were maintained in phenol red–free modified Improved Minimum Essential Medium (IMEM) containing 700 μg/mL G418 sulfate, 1 μmol/L letrozole and 5% charcoal-stripped FBS. AnaR cells were grown in modified IMEM with 700 μg/mL G418 sulfate, 20 μmol/L anastrozole, and 10% charcoal-stripped FBS. The LTLTCa-Nrf2 knock down (LTLTCa-Nrf2KD) cells were cultured in modified IMEM medium supplemented with 700 μg/mL G418 and 5% charcoal-stripped FBS. MCF-7Ca, LTLTCa, AC1, and AnaR cells were grown in monolayer in medium containing 1% penicillin/streptomycin in an incubator at 37°C with 95% air and 5% CO_2.

LTLTCa cells were transduced with Nrf2 shRNA or control shRNA lentiviral particles and cells stably expressing Nrf2 shRNA or control shRNA were selected in the presence of 10 μg/mL puromycin and designated as LTLTCa-Nrf2 knockdown (LTLTCa-Nrf2KD) and LTLTCa-Vector control (LTLTCa-V), respectively.

Cells were untreated or treated with proteasome inhibitors MG-132 or epoxomicin or DMSO vehicle control. The cells were washed with cold PBS and lysed in radioimmunoprecipitation assay (RIPA) buffer (50 mmol/L Tris, pH 8.0, 150 mmol/L NaCl, 0.2 mmol/L EDTA, 1% Nonidet P-40, 0.5% sodium deoxycholate) supplemented with 1× protease inhibitor (Roche Applied Science). Subcellular fractionation of the cells was performed according to manufacturer's protocol (Active Motif). Proteins were quantified using Bio-Rad protein assay. The cell lysates (30–50 μg) were separated on SDS-PAGE and transferred to nitrocellulose membranes. The membranes after blocking in 5% nonfat milk solution in Tris buffered Saline Tween-20 (TBST) were incubated with the primary antibodies overnight at 4°C and washed four times with TBST. This was followed by incubation with secondary antibody at room temperature for 1 hour and washed four times with TBST. The protein bands were visualized using chemiluminescence (ECL) system (Thermo Scientific, product no. 32209). ImageJ software (NIH, Bethesda, MD) was used to quantify the intensity of proteins bands. The protein bands were normalized against loading controls.

Cells were treated with 25 μg/mL cycloheximide for the indicated time points, washed twice with ice-cold 1× PBS, and lysed in RIPA buffer with protease inhibitors. Thirty micrograms of total cell lysate was loaded per well of 10% SDS-PAGE gel, transferred, and immunoblotted with Nrf2 and β-actin antibodies. Nrf2 band intensity was quantified and normalized to β-actin. The relative levels of Nrf2 from sample with zero (0) minute was considered as initial level. The graphs represent the natural logarithm of the relative levels of the Nrf2 protein as a function of the cycloheximide chase time. The half-life of protein was determined in the linear range of the degradation curve.

For ubiquitination assay, cells were treated with 2 μmol/L of MG-132 for 16 hours and lysed in RIPA buffer. One milligram of whole-cell lysate was immunoprecipitated with 1 μg of rabbit IgG or Nrf2 antibody by incubating the reaction mixture overnight in RIPA buffer supplemented with 0.1% SDS at 4°C. After adding 20 μL of washed protein A/G plus beads (Santa Cruz Biotechnology), the mixture was incubated for 2 hours at 4°C and centrifuged at 4,000 rpm for 1 minute. The beads were washed twice with RIPA buffer. Thirty microliters of SDS sample dye was added to each tube and boiled for 5 minutes and immunoprecipitated Nrf2 was separated by 8% SDS-PAGE and immunoblotted with anti-ubiquitin antibody and the same blot was reprobed for Nrf2.

MCF-7Ca, LTLTCa, and LTLTCa-Nrf2KD cells were seeded at the density of 10,000; 20,000 and 20,000 cells per well, respectively, in 24-well plates. After 24-hour incubation, cells were treated with different concentrations of exemestane (viz. 0, 5, 10, 20, and 30 μmol/L) for 72 hours. The cells were incubated with freshly prepared MTT dye (200 μL/well of 5 mg/mL MTT dye in PBS) for 2 hours. MTT dye is reduced by mitochondria aldehyde dehydrogenase to form insoluble formazan crystals. The amount of formazan produced is proportional to viable cells. After dissolving formazan crystals in DMSO, absorbance was recorded spectrophotometrically at 570 nm. Cell viability was calculated from absorbance and normalized to the value of the corresponding vehicle control cells. Each data point represents a mean ± SD from three independent experiments.

ALDEFLUOR assay (Stem Cell Technologies) was performed according to the manufacturer's instructions. MCF-7Ca and LTLTCa cells and LTLTCa-V and LTLTCa-Nrf2KD cells expressing ALDH were stained with Aldefluor reagent and identified by comparing the same sample with and without the ALDH inhibitor diethylaminobenzaldehyde (DEAB). Cells were acquired using FACSCanto and analyzed using FlowJo software (BD Biosciences). Dead cells were excluded on the basis of light scatter characteristics and using viability dye (propidium iodide) gating parameters.

Aldefluor assay/Aldehyde dehydrogenase assay (Stem Cell Technologies) was performed according to the manufacturer's instructions. Briefly, LTLTCa cells were stained with Adlefluor reagent along with the inhibitor of ALDH, DEAB, and sorted using FACS ARIA (BD Biosciences). All cells showing differential ALDH-staining pattern were sorted and designated as ALDH-high and ALDH-low cells based on highest and lowest expression of ALDH enzyme, respectively.

Mammosphere assay was performed using reagents from Stem Cell Technologies, as per manufacturer's instructions. Briefly, LTLTCa-V and LTLTCa-Nrf2KD cells were suspended in complete Mammocult media and 10,000 cells per well were plated in ultra-low attachment 24-well plates. Mammospheres were counted after 3 weeks. Stabilized spheres with a colony count of at least 50 cells were considered as mammospheres (34).

Total RNA was isolated from the untreated cells and cells treated with DMSO or t-BHQ for the indicated time periods using RNeasy mini kit, following manufacturer's protocol. cDNA was synthesized from 1 μg of total RNA as template and the cDNA was used to determine the target gene expression by quantitative real-time PCR using TaqMan gene expression assays.

DCFDA Cellular ROS detection assay kit was used to measure the cellular levels of ROS. Cells were trypsinized and washed with PBS. The cells were suspended in 2′,7′-dichlorofluorescein diacetate (DCFDA) and incubated at 37°C for 30 minutes in the dark and washed with 1× buffer. A total of 10⁶ DCFDA-stained cells were suspended in 1 mL of 1× supplemented buffer and 10⁵ cells in 100 μL of the cell suspension were added to each well of 96-well black plate. A total of 50 μmol/L of t-butyl hydroperoxide was added and the cells were incubated at 37°C for 3 hours to generate ROS as positive control. Using TECAN Infinite M1000 PRO plate reader, ROS-mediated fluorescence intensity was recorded with excitation wavelength at 485 nm and emission wavelength at 535 nm.

Total glutathione content was determined spectrophotometrically using Cayman's glutathione assay kit following the manufacturer's protocol. Briefly, the cells were seeded on 6-well plates on day 1 and harvested on day 3 and lysed in 1× buffer supplied in Glutathione detection kit and oxidized and reduced form of glutathione was quantified following the kit protocol using TECAN plate reader (405 nm). Glutathione content is expressed as μmol/L/μg protein.

Data from cell survival, cell death assay, and real-time PCR were analyzed using a two-tailed Student test. Data were presented as the mean ± SD. Two datasets with P < 0.05 were considered as statistically significant.

Letrozole-sensitive MCF-7Ca and -resistant LTLTCa cells were analyzed for ROS and immunoblotted for Nrf2, INrf2, Nrf2-regulated proteins, and actin (Fig. 1). The results demonstrated that drug-resistant LTLTCa cells contain higher ROS, lower INrf2, and higher Nrf2 levels, as compared with drug-sensitive MCF-7Ca cells (Fig. 1A and B). Subcellular fractionation followed by immunoblotting analysis revealed that nuclear Nrf2 was significantly higher in LTLTCa cells, as compared with MCF-7Ca cells (Fig. 1C). In the same experiment, the cytosolic fraction did not show Nrf2 in either LTLTCa or MCF-7Ca cells (Fig. 1C). The resistant LTLTCa cells also demonstrated significantly increased Nrf2-regulated GCLC (catalytic subunit of glutathione synthesizing enzyme γ-GCS), heme oxygenase-1 (HO-1), drug transporters (MRP-1, MRP-4, and BCRP), and antiapoptotic (Bcl-xL and Mcl-1) proteins, as compared with sensitive MCF-7Ca cells (Fig. 1D). Further analysis of letrozole-sensitive and -resistant cells demonstrated an increase in total and reduced glutathione in resistant LTLTCa cells, as compared with sensitive MCF-7Ca cells (Fig. 1E). In similar experiments, a second cell line AnaR that is resistant to another aromatase inhibitor anastrozole, also showed lower INrf2, higher Nrf2, and GCLC levels, as compared with drug-sensitive AC1 cells (Fig. 1F). In addition, the AnaR cells showed increased expression of Nrf2 downstream genes encoding detoxifying enzymes (GCLC, HO-1), as compared with drug-sensitive AC1 cells (Fig. 1F). Together, these results indicate that persistent treatment of cells with aromatase inhibitors increase ROS, decreases INrf2, increases nuclear Nrf2, increases expression of Nrf2-regulated genes and levels of reduced glutathione, and suggest that Nrf2 and Nrf2-regulated genes play a role in aromatase inhibitor resistance. Interestingly, both letrozole-resistant LTLTCa and anastrozole-resistant AnaR cells containing higher levels of Nrf2 showed downregulation of the Nrf2 downstream gene NQO1, as compared with sensitive cells (Supplementary Fig. S2). The reasons for downregulation of NQO1 gene expression in aromatase inhibitor–resistant cells remain unknown. It is noteworthy that the lack of induction of NQO1 gene in letrozole-treated Hepa 1c1c7 cells was observed earlier (35).

LTLTCa cells were transduced with either lentiviral vector (control) or Nrf2shRNA viral particles and positive clones selected in puromycin. MCF-7Ca, LTLTCa-V (vector control), and LTLTCa-Nrf2KD (Nrf2 knockdown) cells were immunoblotted for Nrf2 and INrf2; detoxifying proteins GCLC and NQO1; membrane transporters MRP1, MRP4, and BCRP; and antiapoptotic proteins Mcl-1, Bcl-xL, and actin (Fig. 2A and B). shRNA silencing of Nrf2 significantly reduced the levels of Nrf2, GCLC, NQO1, MRP4, Bcl-xL, and Mcl-1 (Fig 2A and B). Nrf2KD cells also showed downregulation of MRP1 but the change was insignificant. This is presumably due to relatively lower contribution of Nrf2, as compared with other factors including NF-κB and c-Jun that regulate expression of MRP1 in LTLTCa cells (15). Previous studies have suggested the option of using steroidal aromatase inhibitor exemestane to treat HER2-negative, hormonal receptor–positive, postmenopausal metastatic breast cancer patients with resistance to nonsteroidal aromatase inhibitor (reviewed in ref. 36). Therefore, one of the aims of the experiment was to evaluate exemestane sensitivity of aromatase inhibitor–resistant LTLTCa cells. The MCF-7Ca, LTLTCa, and LTLTCa-Nrf2KD cells were compared for exemestane sensitivity (Fig. 2C). Interestingly, the treatment of LTLTCa cells (expressing higher Nrf2 compared with MCF-7Ca cells) with exemestane showed some degree of sensitivity that increased with increasing concentration of exemestane (Fig. 2C). However, MCF-7Ca cells containing lower Nrf2 showed significant sensitivity to 20 and 30 μmol/L exemestane, as compared with LTLTCa cells containing higher Nrf2 (Fig. 2C). Intriguingly, shRNA inhibition of Nrf2 significantly sensitized LTLTCa-Nrf2KD cells to exemestane (Fig. 2C). The 20 and 30 μmol/L exemestane concentrations significantly decreased cell survival in LTLTCa-Nrf2KD cells as compared with LTLTCa cells (Fig. 2C). Furthermore, the Nrf2 levels in LTLTCa-Nrf2KD cells were similar to MCF-7Ca cells (Fig. 2A) and their sensitivities to exemestane were not significantly different (Fig. 2C; P > 0.7758). In other words, knockdown of Nrf2 in LTLTCa-Nrf2KD cells sensitized cells to exemestane to a similar extent as observed with MCF-7Ca cells. Interestingly, LTLTCa-Nrf2KD cells also showed increased sensitivity to genotoxic antitumor drugs doxorubicin and etoposide as compared with LTLTCa cells (Supplementary Fig. S3). Together, these results suggested a role for Nrf2 in aromatase inhibitor drug resistance. It is noteworthy that LTLTCa cells contained significantly lower ERα, as compared with MCF-7Ca cells and shRNA inhibition of Nrf2 in LTLTCa cells had more or less no effect on ERα level in LTLTCa cells (Supplementary Fig. S4), indicating that Nrf2 does not regulate ERα expression in resistant cells.

LTLTCa cells demonstrated decreased Nrf2 ubiquitination and degradation, as compared with MCF-7Ca cells (Fig. 3A). In related experiments, the rate of degradation of Nrf2 was significantly lower in LTLTCa cells, compared with MCF-7Ca cells (Fig. 3B). These results collectively suggested that higher levels of Nrf2 in LTLTCa cells are due to decreased ubiquitination and degradation of Nrf2. Notably INrf2, which functions as an adaptor protein for Cul3-Rbx1–mediated ubiquitination and degradation of Nrf2, is downregulated in aromatase inhibitor–resistant LTLTCa cells (Figs. 1B and 3C). Therefore, it is reasonable to conclude that lower INrf2 levels were responsible for the reduced ubiquitination and degradation of Nrf2 in LTLTCa cells. We also determined whether downregulation of INrf2 in LTLTCa cells is due to degradation and/or decreased transcript levels of the INrf2 gene, as compared with MCF-7Ca cells (Fig. 3C–E). The treatment of LTLTCa cells with proteasome inhibitors MG132 (Fig. 3C) or epoxomicin (Fig. 3D) failed to stabilize INrf2 indicating that the lower INrf2 level in LTLTCa cells is not due to degradation of INrf2. We also performed experiments to investigate whether epigenetic and/or autophagy mechanisms contribute to lower levels of INrf2 in LTLTCa cells. Epigenetic regulation of INrf2 (Keap1) was investigated by treating LTLTCa cells with an inhibitor of DNA methyltransferase and inhibitors of histone deacetylase. Treatment with those epigenetic modulators failed to restore the levels of INrf2. We also utilized inhibitors of autophagy to examine the possibility of INrf2 degradation by autophagy (37). Even though we observed higher level of autophagy in LTLTCa cells as compared with MCF-7Ca cells, the inhibition of autophagy did not increase the levels of INrf2. It is noteworthy that negative data on epigenetic and autophagy regulation of INrf2 are not included. Interestingly, RT-PCR analysis of INrf2 RNA demonstrated significantly lower INrf2 RNA transcripts in aromatase inhibitor–resistant LTLTCa cells, as compared with drug-sensitive MCF-7Ca cells (Fig. 3E). This result suggested that INrf2 gene expression is downregulated in aromatase inhibitor–resistant cells. RT-PCR analysis also showed a marginal increase in Nrf2 gene expression in LTLTCa cells, as compared with sensitive MCF-7Ca cells that might also have contributed to higher Nrf2 in resistant cells (Supplementary Fig. S3).

Recent studies have implicated mammary TICs in resistance to chemotherapy and radiation (38, 39). TICs are immature, poorly differentiated, and highly tumorigenic (40–42). TICs have a decreased ability to undergo apoptosis and a higher ability for DNA repair, making them more resistant to cancer therapy, compared with differentiated counterparts (43, 44). We isolated TICs expressing ALDH from LTLTCa cell culture. It has been reported that chemoresistant cancer stem cells have high ALDH activity (45) and ALDH is considered as a marker of normal and malignant human mammary stem cells (46). MCF-7Ca, LTLTCa, LTLTCa-low ALDH, and LTLTCa-high ALDH cells were lysed and immunoblotted for Nrf2, INrf2, GCLC, DNA repair proteins, and HER2 (Fig. 4). Results revealed that TICs expressed lower levels of INrf2 and higher levels of Nrf2 and Nrf2 downstream GCLC gene expression. TICs also expressed higher levels of HER2. Intriguingly, TICs with high ALDH levels (stem cells) expressed significantly higher levels of nonhomologous end-joining (NHEJ) DNA repair proteins Ku80 and Ku70, as compared with MCF-7Ca and LTLT-Ca cells. This observation has high significance as TICs are believed to contribute to drug resistance.

MCF-7Ca and LTLTCa cells were immunoblotted for Nrf2 (Fig. 5A) and in separate experiments were stained to assess the expression of stem cell marker ALDH in the absence and presence of ALDH inhibitor DEAB (Fig. 5B). As expected, the levels of Nrf2 and ALDH were significantly higher in LTLTCa cells compared with MCF-7Ca cells (Fig. 5A and B). This indicated the presence of an increased stem cell-like population in LTLTCa cells containing higher levels of Nrf2, as compared with MCF-7Ca cells with lower Nrf2 protein. In related experiments, LTLTCa-V (vector control), LTLTCa-Nrf2KD clone #2 and clone #1 cells were immunoblotted for Nrf2 and actin (Fig. 5C). The results demonstrated that LTLTCa-V cells showed highest expression of Nrf2, which was followed by clone #2 and clone #1 cells LTLTCa-V and the two clones of LTLTCa-Nrf2KD cells were stained for assessing the expression of stem cell marker ALDH in the absence and presence of ALDH inhibitor DEAB (Fig. 5D). Results demonstrated a direct correlation between Nrf2 expression levels and the magnitude of ALDH expression. LTLTCa-V cells expressing the highest levels of Nrf2 led the highest percentage of cells in a gated region R3 that stained positive for ALDH. Moreover, clone #2, containing lower expression levels of Nrf2 demonstrated significantly decreased ALDH. Notably, shRNA downregulation of Nrf2 in LTLTCa (clone #1) containing lowest level of Nrf2 also showed the least staining for ALDH. These results showed that shRNA inhibition of Nrf2 led to an Nrf2-dependent decrease in ALDH-positive TICs. It is noteworthy that ALDH is an Nrf2 downstream gene and its expression is regulated by Nrf2 (47–49). Therefore, our observation of a relationship between Nrf2 and ALDH is strengthened by previous reports (47–49). In related experiments, LTLTCa and both clones of LTLTCa-Nrf2KD cells were also analyzed for mammosphere formation (Fig. 6). The results (compare Figs. 5C and 6) revealed a direct correlation between Nrf2 and mammosphere formation. shRNA inhibition of Nrf2 in LTLTCa cells led to Nrf2 concentration–dependent decrease in mammosphere formation. Together, the results suggest a direct correlation between Nrf2, ALDH-positive TICs and mammosphere formation with implications in aromatase inhibitor drug resistance that warrant further studies.

This is the first report demonstrating a role for Nrf2 in aromatase inhibitor resistance in breast cancer. Letrozole-resistant LTLTCa cells generated significantly higher ROS levels, as compared with letrozole-sensitive MCF-7Ca cells. This increase in ROS was more significant considering that reduced glutathione that scavenges ROS was also increased. We believe that long-term treatment of letrozole could result in continuous generation of ROS, which is known to activate Nrf2. However, Nrf2-mediated antioxidant gene expression might not be sufficient to lower the levels of ROS leading to higher levels of ROS in drug-resistant LTLTCa cells, as compared with drug-sensitive MCF-7Ca cells. LTLTCa cells also showed lower INrf2 and higher expression levels of Nrf2, as compared with MCF-7Ca cells. Similar results were also observed in anastrozole-resistant AnaR cells. The increase in ROS and decrease in INrf2 led to decreased ubiquitination and degradation of Nrf2 leading to the stabilization and nuclear translocation of Nrf2. High levels of Nrf2 in the nucleus led to coordinated induction of Nrf2 downstream cytoprotective proteins including detoxifying enzymes, membrane transporters, and antiapoptotic proteins. On the basis of the documented role of Nrf2-activated cytoprotective proteins in drug resistance in other systems (28), our results strongly suggest that Nrf2 plays a role in aromatase inhibitor resistance. This was further supported by studies showing that inhibition of Nrf2-sensitized LTLTCa cells to the aromatase inhibitor exemestane.

Our studies also revealed that lower levels of INrf2 in drug-resistant LTLTCa cells is not due to instability of INrf2 protein as inhibitors of proteasomes failed to increase the level of INrf2. Furthermore, the lower levels of INrf2 in LTLTCa cells are also not due to epigenetic modulation or autophagy as inhibitors of DNA-methyl transferase, histone deacetylation, and autophagy all failed to increase INrf2 levels in aromatase inhibitor–resistant LTLTCa cells. Additional studies demonstrated that downregulation of INrf2 in letrozole-resistant LTLTCa cells is due to a significant decrease in INrf2 transcripts. This raises an intriguing question regarding the mechanism of aromatase inhibitor–mediated downregulation of INrf2 gene expression and is a subject of future studies.

Aromatase inhibitor–resistant cells showed higher population of TIC cells that are believed to contribute to resistance to chemotherapy and radiation (34, 39). The TICs showed lower INrf2 and higher Nrf2. TICs also showed higher expression of Nrf2 downstream cytoprotective proteins and higher levels of Ku70 and Ku80 proteins that participate in the NHEJ pathway that is known to be active in repairing DNA double-strand breaks, one of the most lethal forms of DNA damage. NHEJ is known to be active in G₀–G₁ phase of the cell cycle in which many “stem” cells reside (50). These observations suggested that higher Nrf2 in TICs contributed to aromatase inhibitor drug resistance. While a previous publication found Ku proteins to be reduced in LTLTCa cells (51), their high expression levels in TICs are significant. NHEJ may participate in drug resistance in “stem”-like TICs by increasing repair of DNA damage, leading to cell survival. This conclusion was further strengthened from following observations. First, shRNA inhibition of Nrf2 led to Nrf2 concentration–dependent reduced survival of TICs. In other words, Nrf2 enhances survival of TICs in the presence of drugs that contribute to resistance. It is noteworthy that a role of Nrf2 in cell survival by inhibiting apoptosis is well established in other systems (16). Second, in related experiments, inhibition of Nrf2 led to significant decrease in mammosphere formation. This signifies the importance of Nrf2 in mammosphere development. Therefore, it is reasonable to conclude that higher Nrf2 levels in TICs contributed to aromatase inhibitor resistance.

Collectively, the results led to the hypothesis (Supplementary Fig. S5) that aromatase inhibitor downregulates INrf2 transcripts that, combined with higher ROS, leads to increased Nrf2 and Nrf2 downstream cytoprotective proteins including detoxifying enzymes, antioxidants, and efflux pumps. In addition, higher Nrf2-mediated increased expression of antiapoptotic proteins that led to reduced apoptosis. Finally, higher Nrf2 increased TIC survival and mammosphere formation. Together, all these Nrf2-dependent processes contributed to aromatase inhibitor drug resistance.

Previous studies have shown increased signaling through HER2 receptors and increased MAPK activity as probable causes of endocrine drug resistance (reviewed in refs. 7 and 52). In addition, breast cancer cells receiving long-term antiestrogen treatment appear to have increased ROS and disruption of reversible redox signaling that involves redox-sensitive factors including protein phosphatases, protein kinases such as ERK, and transcription factors AP-1 and NF-κB that contribute to drug resistance (reviewed in ref. 53). Furthermore, PI3K–Akt–mTOR pathway has also been implicated in endocrine drug resistance (54). The INrf2:Nrf2 system identified in the current report is a novel mechanism of aromatase inhibitor drug resistance but is also related to the abovementioned mechanisms. This is because Nrf2 is a transcription factor that controls redox homeostasis (reviewed in ref. 12). In addition, Nrf2 itself is regulated by PI3K–Akt–mTOR, MAPKs, and redox-sensitive factors including AP1 (reviewed in ref. 12). Further studies are required to explore the exact relationship between these factors and pathways leading to aromatase inhibitor drug resistance and possible therapeutic intervention.

Recent studies have shown some benefit in using steroidal aromatase inhibitor exemestane in treating HER2-negative, hormonal receptor–positive, and postmenopausal metastatic breast cancer patients with resistance to nonsteroidal aromatase inhibitor (36). The studies in this report suggest that it might be possible to increase the efficacy of exemestane in the patients with endocrine drug resistance by inhibiting Nrf2. This assumption is based on observation that shRNA inhibition of Nrf2 in letrozole-resistant cells significantly increased the sensitivity to exemestane.

In conclusion, the current studies present strong evidence for a role of INrf2:Nrf2 in aromatase inhibitor drug resistance in breast cancer. The mechanism(s) involving Nrf2 to combat drug resistance is especially interesting as it is estrogen independent. The studies also suggest that Nrf2 inhibitors from natural sources could be explored for use as adjuvants with aromatase inhibitor drugs to treat aromatase inhibitor–resistant breast cancer.

No potential conflicts of interest were disclosed.

Conception and design: R. Khatri, R. Guha, A.E. Tomkinson, A. Brodie, A.K. Jaiswal

Development of methodology: P. Shah, A. Brodie

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Shah, R. Guha, A. Brodie

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Khatri, R. Guha, A. Brodie, A.K. Jaiswal

Writing, review, and/or revision of the manuscript: R. Khatri, P. Shah, R. Guha, F.V. Rassool, A.E. Tomkinson, A. Brodie, A.K. Jaiswal

Study supervision: A.K. Jaiswal

The authors thank their colleagues at the University of Maryland School of Medicine (Baltimore, MD) for helpful discussions.

This work was supported by NIH grant RO1 ES012265 (to A.K. Jaiswal), RO1 ES021483 (to A.K. Jaiswal), RO1 CA62483 (to A. Brodie), RO1 GM047466 (to A.K. Jaiswal), and a grant from V-Foundation and Endow Funds (to A.K. Jaiswal).

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.