Introduction

Selective estrogen receptor modulators (SERMs) are clinically important for the treatment and prevention of breast cancer as well as for therapy of postmenopausal osteoporosis1. The first generation triphenylethylene SERM, tamoxifen, significantly decreases the incidence of breast cancer in high-risk women and has good efficacy as a first line of defense in breast cancer therapy2. However, its associated side effects such as an increased incidence of endometrial cancer limit the long-term usage of this drug1,2.

Benzothiophene SERMs, including raloxifene and arzoxifene (Figure 1), have demonstrated a more favorable therapeutic and safety profile in clinical as well as preclinical studies1. Raloxifene is a second generation SERM that has been widely used for the prevention and treatment of postmenopausal osteoporosis and for breast cancer chemoprevention in high-risk women3,4. Arzoxifene comes from a new generation of benzothiophene SERMs5, and both arzoxifene and its active metabolite desmethylarzoxifene (DMA) have a higher antiestrogenic potency than tamoxifen and better bioavailability than raloxifene6. Due to the absence of agonist activity in the endometrium, the presence of antiestrogenicity in the breast and estrogenicity in bones1, it is believed that benzothiophene SERMs are safer for long-term treatment compared to tamoxifen. Currently, there are no reports suggesting that raloxifene and arzoxifene could cause DNA damage and carcinogenesis. In contrast, it has been reported that reactive quinoid metabolites formed from benzothiophene SERMs are capable of inducing cytoprotective quinone reductase (NQO1) activity in both cell and animal models7.

Figure 1
figure 1

Structures of benzothiophene compounds. (A) Chemical structures of the benzothiophene SERMs raloxifene, arzoxifene, and desmethylarzoxifene, the benzothiophene compound BC-1 and its analogues BC-2 and BC-3. (B) Formation of di-quinone methide from benzothiophene compounds.

PowerPoint slide

Induction of cytoprotective enzymes is largely mediated by the antioxidant responsive element (ARE), which promotes the transcription of many antioxidant genes such as NQO1, superoxide dismutase isoenzymes (SOD) and glutathione S-transferase (GST)8,9. The cellular sensor protein Kelch-like ECH-associated protein 1 (Keap1) contains large numbers of sulfhydryl groups that can be modified by oxidants and electrophiles10,11. Under basal conditions, Keap1 associates with the key transcriptional factor nuclear factor erythroid 2-related factor 2 (Nrf2) and targets it for degradation12. When Keap1 is modified by either oxidative or electrophilic reagents, proteolysis is inhibited, and the dissociated Nrf2 translocates to the nucleus, binds to ARE, and promotes phase II enzyme gene expression12. Therefore, oxidants and electrophiles might initiate the inducer signal by modifying Keap1 sulfhydryl groups, which can be further transduced by the Keap1-Nrf2 signaling pathway to activate cellular ARE.

All SERMs and their active metabolites share common polyaromatic scaffolds that are able to form quinoids13. These quinoid metabolites are reactive electrophiles that can covalently modify cellular macromolecules and either cause genotoxicity and cytotoxicity or contribute to cytoprotection14. It has been reported that in the rat liver and intestine, tamoxifen could induce the phase II enzymes aryl sulfotransferase and hydroxysteroid sulfotransferase15. Separate studies reported that 4-hydroxytamoxifen and raloxifene could induce NQO1 in MCF-7 cells via an estrogen receptor (ER) dependent mechanism16,17. In our previous work, it was found that arzoxifene and its active metabolite DMA could induce NQO1 independently of ER pathways in both cells and rat models7. These data suggest that SERMs might produce beneficial quinoid electrophiles and activate the Keap1-Nrf2-ARE pathways. Further studies showed that a benzothiophene core compound could significantly induce NQO1 activity and ARE signaling compared to arzoxifene and DMA7, which might serve as a potential benzothiophene SERM for chemoprevention. However, the detailed mechanism of its antioxidant signaling induction is still unclear.

In this study, the potential mechanism of ARE induction by the compound BC-1 was investigated based on the hypothesis that its reactive metabolites could modify Keap1 and thus activate ARE. Using cell models, the induction capability of the ARE signaling pathway by BC-1 was measured, the possible regulatory factors (including ER and Nrf2) were examined, and the in vitro quinoid formation of BC-1 was analyzed. Compared with two analogues of BC-1 that can not form reactive di-quinone methides (Figure 1), it was assumed that this quinoid formation of BC-1 played a key role in ARE induction via Keap1-Nrf2 and, therefore, suggested a novel mechanism of antioxidant signaling activation by benzothiophene SERMs via reactive metabolite formation.

Materials and methods

Reagents

All chemicals and reagents were purchased from Sigma-Aldrich (St Louis, MO, USA) unless stated otherwise. Cell culture media and supplements were obtained from Life Technologies (Grand Island, NY, USA). The luciferase substrate was obtained from Promega (Madison, WI, USA). The BCA assay kit was obtained from Thermo Fisher Scientific (Rockford, IL, USA). The benzothiophene compounds BC-1, BC-2, BC-3 were synthesized by J&K Scientific Ltd (Beijing, China) as described in the literature7.

Cell culture

Hepa1c1c7 murine hepatoma cells were cultured in α-MEM with 1% penicillin-streptomycin and 10% fetal bovine serum (FBS). HepG2 cells stably transfected with ARE-luciferase reporter were cultured in F-12 medium with 10% FBS, 1% penicillin-streptomycin, 1% essential amino acids, and 0.2 mg/mL insulin. All cells were incubated in 5% CO2 at 37 °C, and the medium was replaced every 2–3 d.

ARE-luciferase reporter assay

HepG2 cells stably transfected with ARE-luciferase reporter were plated in a 24-well plate (2×104 cells per well), and cells were treated with either control (DMSO) or test compounds and incubated for various lengths of time. The luciferase activity was determined according to the protocol provided by the manufacturer (Promega). Briefly, cells were washed with cold phosphate buffered saline (PBS) twice and lysed with 500 μL of reporter lysis buffer. After centrifugation, 20 μL of the supernatant was used for measurement of the luciferase activity and normalized to its protein concentration using the BCA assay. The data were obtained from three separate experiments and expressed as fold-induction over the control.

siRNA interference

Small interfering RNA (siRNA) experiments were performed on HepG2-ARE cells using anti-Nrf2 and nonspecific scramble siRNA lentiviral transfection particles (Sigma). NM_006164.2-1144s1c1 (5′-CCGGCCGGCATTTCACTAAACACAACTCGAGTTGTGTTTAGTGAAATGCCGGTTTTT-3′) was selected based on preliminary data. HepG2 cells stably expressing ARE luciferase report were plated in 96-well plates (0.5×104 cells per well) one day prior to transfection. On d 2 when the cell confluency reached 70%–80%, the cell culture medium was changed, and 150 μL fresh medium containing hexadimethrine bromide (4 mg/mL stock solution in ddH2O, final concentration 8 μg/mL) was added to the cells and gently swirled in the plate to mix. Lentiviral particles were thawed at room temperature and added to cells, and plates were gently swirled to mix. Plates were centrifuged at 500×g for 1 h at room temperature and incubated at 37 °C for 16 h. The cell culture medium was changed on d 3, and on d 4, the compounds (5 μmol/L) were added to the cells, and the luciferase assay was performed after a 6 h incubation.

Rat liver microsomal incubation

A solution containing the test compounds, rat liver microsomes (1.0 mg/mL), reduced L-glutathione (GSH, 500 μmol/L) and reduced β-Nicotinamide adenine dinucleotide 2'-phosphate (NADPH, 1 mmol/L) in 50 mmol/L phosphate buffer (pH 7.4, 500 μL total volume) was incubated for 30 min at 37 °C. For comparisons, either GSH or NADPH were omitted as controls. The reactions were terminated by chilling in an ice bath followed by the addition of methanol (500 μL) containing perchloric acid (50 μL/mL). The reaction mixtures were centrifuged at 15 000×g for 10 min, and aliquots (80 μL) of the supernatant were analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS).

In vitro metabolites identification

Analytical liquid chromatography-ultraviolet (LC-UV) analysis was performed using a 5 μmol/L, 4.6 mm×250 mm Aglient Eclipse XDB C18 column at a flow rate of 1.0 mL/min. The mobile phase A consisted of H2O with 0.1% formic acid, and the mobile phase B was acetonitrile with 0.1% formic acid. The initial mobile phase consisted of 10% B followed by a 40 min linear gradient to 30% B, then a linear gradient from 30% to 60% B for 10 min, and finally a 5 min gradient from 60% to 90% B. The LC-MS/MS spectra were obtained using a Shimadzu LCMS-IT-TOF mass spectrometer (Kyoto, Japan).

Statistical analysis

Student's t-test and one-way ANOVA analysis with Tukey's multiple comparison test were performed using GraphPad Prism version 4.00 for Windows (La Jolla, CA, USA).

Results

Estrogen receptor independent mechanism for ARE activation in cells

The potential of BC-1 to activate ARE was comparable to 4-bromoflavone, which is an effective chemopreventive reagent, as observed in our previous study7. The data from our pilot study also showed that a dose of BC-1 between 1 μmol/L and 20 μmol/L could activate ARE (unpublished data). Because 5 μmol/L of BC-1 could strongly activate ARE but was below its IC50 (26 μmol/L), this concentration was applied to all cellular experiments in this study.

The roles of ERs, especially ERβ, in the SERM-induced phase II enzyme induction are controversial16,17. To investigate the role of ERs in BC-1 induced ARE activation, HepG2-ARE cells were pretreated with either the ER agonist E2 or antagonist ICI 182,780 prior to the addition of BC-1. ARE induction was examined in cells that were treated with vehicle control, E2, ICI 182,780, and BC-1 individually, as well as cells that were treated with various combinations of these compounds. The results indicated that E2, ICI 182,780, and vehicle DMSO could not induce ARE activation, and BC-1 was the only compound that could induce activity alone (Figure 2). BC-1 was a potent inducer of ARE-luciferase activity, with a 5 μmol/L dose resulting in a 17-fold increase compared to the control. Co-treatment of BC-1 with either E2 or ICI 182,780 neither inhibited nor enhanced BC-1 induced ARE activation significantly (Figure 2). Therefore, it is suggested that ERs are not involved in BC-1 induced ARE activation in HepG2 cells.

Figure 2
figure 2

The effects of an ER agonist and antagonist on BC-1 induced ARE luciferase activity. HepG2-ARE cells were treated with vehicle, the ER agonist E2 (5 μmol/L), the ER antagonist ICI 182,780 (5 μmol/L), BC-1 (5 μmol/L), and various combinations of these compounds. Luciferase activity was measured after treatment for 24 h. The results were expressed as the mean±SD and compared using one-way ANOVA with Tukey's post-test. cP<0.01 compared to the controls.

PowerPoint slide

Involvement of Nrf2 in the ARE induction by BC-1

HepG2-ARE cells were further transduced with siRNA lentiviral particles targeting Nrf2 to examine the role of Nrf2 in ARE activation by BC-1. The results showed that the transfection of anti-Nrf2 siRNA significantly decreased BC-1 induced ARE luciferase activity compared to the cells transduced with control siRNA, whereas no difference was observed in the cells receiving DMSO treatment (Figure 3).

Figure 3
figure 3

Effects of anti-Nrf2 siRNA on BC-1 induced ARE luciferase activity. HepG2-ARE cells were treated with either DMSO or 5 μmol/L BC-1 and transfeced with either anti-Nrf2 or non-specific scrambled siRNA (NSC siRNA) lentiviral particles. Luciferase activity was measured at 24 h post transfection, and comparisons were made between the controls and the treated cells. cP<0.01 compared between the different treatments.

PowerPoint slide

Other factors such as mitogen-activated protein kinase (MAPK) and protein kinase C (PKC) are known to be involved in ARE activation18. Using specific inhibitors to MAPK and PKC (U0126 and staurosporine, respectively), it was observed that inhibition of either kinase had no effect on BC-1 induced ARE activity (unpublished data). Therefore, whether these kinases play a role in BC-1 induced ARE deserves future investigation.

Quinoid formation of BC-1 in rat liver microsomes

The benzothiophene core compound BC-1 contains a polyaromatic phenolic scaffold, which can be bioactivated into quinoids. To investigate whether BC-1 can generate quinoid metabolites, rat liver microsomes either with or without GSH were incubated with 10 μmol/L of BC-1. Unstable reactive quinoid intermediates formed in incubations were trapped with GSH, and then extracted and analyzed by LC-UV and LC-MS/MS (Figure 4 and Supplemental Figures 1–3).

Figure 4
figure 4

Quinone formation of BC-1 in rat liver microsomes. LC-UV chromatogram of BC-1 (10 μmol/L) incubated with rat liver microsomes in the presence (A) and absence (B) of GSH.

PowerPoint slide

In the LC-UV chromatogram, there were two major types of quinoids: di-quinone methide and o-quinone, both of which were identified as reactive metabolites of BC-1 in microsomal incubations (Figure 4A). These quinoids went on to form di-quinone methide GSH conjugates, o-quinone GSH conjugates, and o-quinone di-GSH conjugates, all of which were identified by LC-MS/MS (Supplemental Figures 1–3).

The major GSH conjugates, BC-1 di-quinone methide monoGSH conjugates, were identified based on the detection of [M+H]+ at m/z 548 at 24.5 min (Supplemental Figure 1). Subsequent collision-induced dissociation (CID) of this ion produced the most abundant ions at m/z 419, which corresponds to the neutral loss of anhydroglutamic acid (129 Da) from GSH. Other fragment ions included those at m/z 273 and m/z 473, which were derived from the cleavage of S-CH2 bond within the cysteine residue (275 Da) and the neutral loss of glycine (75 Da) from parent compound, respectively. These fragmentations have been considered to be the unique CID fragmentations of GSH for confirmation of the formation of drug-GSH conjugates19,20.

BC-1 also formed o-quinone mono GSH conjugates based on the detection of its protonated ions at m/z 564 at 21.0 min. CID of this ion also gave similar fragmentation derived from GSH as the di-quinone methide GSH conjugates (Supplemental Figure 2). The base peak at m/z 435 resulted from the neutral loss of anhydroglutamic acid, the ion at m/z 418 from the neutral loss of glutamine (146 Da), and the ion at m/z 343 from the loss of their combination. The ion at m/z 289 was derived from the cleavage of S-CH2 bond within the cysteine residue.

BC-1 o-quinone di-GSH conjugates were identified based on the detection of [M+2H]2+ at m/z 435 at 15.0 min (Supplemental Figure 3). CID fragmentation of this ion also gave similar fragment ions derived from the loss of different moieties of GSH.

Thus, BC-1 could largely form reactive quinoids in microsomal incubations, which were trapped by GSH. The formation of similar reactive intermediates from BC-1 in cells and in vivo is possible, which might contribute to ARE activation via reaction with cellular macromolecules. However, the exact structures of the formed conjugates are unclear because BC-1-derived quinoids could form several GSH conjugates, including isomers. Detailed investigations on BC-1's GSH conjugates may be warranted in the future.

Modulation of ARE induction activity by BC-1 analogues

To modulate the antioxidant activity of BC-1, a series of analogues were synthesized, in which the hydroxyl group at the 4′ position of BC-1 was replaced with either hydrogen (BC-2) or fluorine (BC-3) (Figure 1). Both of the analogues had comparable estrogenic activity with BC-1 in cells (Supplemental Table 1); however, BC-2 and BC-3 could not form di-quinone methides (Figure 1).

Induction of ARE activation in HepG2 cells by BC-1 and its analogues at 5 μmol/L were measured (Figure 5). As a potent phase II enzyme inducer, BC-1 activated ARE approximately 15-fold compared to the control; however, both BC-2 and BC-3 showed no significant activation of ARE. Therefore, these data suggested a correlation between ARE activation by this benzothiophene compound and its ability to form reactive di-quinone methides.

Figure 5
figure 5

Induction of ARE signaling pathways by BC-1 analogues. HepG2-ARE cells were incubation with either BC-1 (5 μmol/L) or one of its analogues (BC-2 and BC-3, 5 μmol/L). Luciferase activity was measured after treatment for 24 h and compared to vehicle control (DMSO). cP<0.01 compared to controls.

PowerPoint slide

Discussion

The predicted long-term clinical use of SERMs and their risk of carcinogenesis require a complete understanding of their detrimental and beneficial effects in humans. Benzothiophene SERMs and their active metabolites contain polyaromatic phenol structures that can form quinoids14,21,22,23,24, which can covalently modify cellular macromolecules and cause different biological consequences13. In this study, we found that reactive metabolites were formed from compound BC-1, which is the core structure of benzothiophene SERMs (Figure 1), and activated ARE signaling pathways via reactive electrophilic metabolites (Figures 1,2,3,4,5). These data strongly support the concept that appropriate benzothiophene compounds can act as chemopreventive agents via the Keap1-Nrf2-ARE pathway and allow for the correlation of benzothiophene compound structure with the potential to induce chemopreventive mechanisms.

A previous study reported that the benzothiophene core compound BC-1 could greatly activate ARE signaling7, but a detailed study on the mechanism of its activation had not been performed. Our data supported the theory that the induction of ARE by BC-1 was ER independent (Figure 2) despite BC-1 being an effective ER modulator (Supplemental Table 1). Interestingly, the chemopreventive effects of the benzothiophene SERM arzoxifene toward ER negative breast cancer have been reported by the Sporn group25. Their data demonstrated that a single treatment of arzoxifene significantly delayed the development of ER negative tumors in a transgenic mouse model25, which suggests that the chemopreventive effect of arzoxifene in this breast cancer model was independent of the ER signaling pathway. Though it is uncertain if the induction of antioxidant signaling by arzoxifene contributes to its anticarcinogenic effects, these data suggested that the beneficial biological effects of benzothiophene compounds may involve mechanisms other than ER pathways.

This study showed that the activation of Keap1 was demonstrated to be important for the ARE induction by BC-1 (Figure 3). ARE activation is one of the most important chemopreventive mechanisms; many inducible antioxidant enzymes such as NQO1, GST, and SOD contain an ARE structure and provide cytoprotection by the detoxification of small molecules8,9. Many chemopreventive agents such as sulforaphane and Michael acceptors have been shown to chemically modify the cysteine residues of the redox sensor protein Keap1 and lead to the activation of ARE by Nrf218,26. Animal studies using Keap1 and Nrf2 knockouts indicated the importance of the Keap1-Nrf2-ARE pathway for antioxidant enzyme induction by sulforaphane and other agents27. In this study, Nrf2 was also shown to be involved in ARE activation by BC-1 (Figure 3), suggesting that the sensor protein Keap1, which regulates Nrf2 translocation, is modified. Further research is required to differentiate the exact mechanisms invoked by BC-1, including the roles of the MAPK and PKC pathways.

The benzothiophene compound BC-1 could form large amounts of di-quinone methide metabolites that are trapped by GSH when incubated with rat liver microsomes, as shown in this study (Figure 4, Supplemental Figures 1–3). These quinoids are reactive electrophiles and are potential oxidation/alkylation agents for biomolecules such as proteins, DNA, and GSH. There is controversy over the degree of specificity of covalent modification by electrophiles, but the study by the Liebler group suggested that the protein modifications were not random and certain protein families were particularly susceptible to alkylation28. It has been reported that quinoids formed from benzothiophene SERMs (including raloxifene and arzoxifene) participate in enzyme inhibition and protein modification23,29,30,31,32. For instance, irreversible inhibition of the cytochrome P450 enzyme CYP3A4 and alkylation of proteins in rat liver microsomes by raloxifene were observed23,29. Recent studies reported a modification on either Cys239 or Tyr75 of CYP3A4 by raloxifene30,31,32, which may play critical roles in the time-dependent inactivation of CYP3A4 by raloxifene. Therefore, after its bioactivation to form di-quinone methides, BC-1 might be responsible for the reactivity towards the Keap1 sensor required to induce ARE.

We elucidated that the formation of reactive di-quinone methides of benzothiophene compounds might play the central role in ARE activation. Benzothiophene SERMs were able to form several types of quinoids, and di-quinone methides were the most abundant electrophilic metabolites formed from benzothiophene compounds13. In vitro, the half-life of the DMA di-quinone methide is approximately 15 s22,33, and the half-life of BC-1 di-quinone methide is approximately 20 s (unpublished data). According to Thompson et al, quinoids with half-lives in the range of 10 s to 10 min have time to diffuse away from the site of formation to react with cellular nucleophiles34. Di-quinone methides prefer to react with either sulfur nucleophiles such as GSH or protein cysteine sulfhydryl groups, and may serve as the chemical trigger for ARE activation. The importance of di-quinone methide formation from BC-1 for its ARE induction was further proven by comparison with its analogues (Figures 1 and 5). When the 4′-OH group of BC-1 was replaced by hydrogen and fluoride (BC-2 and BC-3, respectively), oxidation of the compounds to form a di-quinoid methide was blocked21, and ARE induction in the cells was completely lost in both BC-2 and BC-3 treatments (Figure 5). Therefore, the formation of reactive di-quinoid methides from BC-1 might be responsible for ARE induction, and preventing their formation would result in the loss of its putative beneficial effects via induction of antioxidant signaling.

Though BC-1 represented the major aromatic structure of raloxifene and arzoxifene, it was a much stronger inducer of ARE signaling compared to both analogues7. A potential explanation for the weak induction by arzoxifene and raloxifene is that direct interaction of the quinoid with the sensor protein is required. The steric hindrance due to the SERM side arm in both raloxifene and arzoxifene would interfere with the interaction of drug metabolites and the sensor proteins (Figure 1). It is also known that raloxifene could form catechols, o-quinones and di-quinone methods, but the half-life of the raloxifene di-quinone methide is only 1 s33. Therefore, the short half-life of raloxifene di-quinone methide may be another explanation for its weak ARE induction. Because the characteristics of quinoid metabolites formed from SERMs are important, engineering a SERM structure to modulate redox activity might influence its biological activity with potential therapeutic significance.

In conclusion, this study showed that the core compound of benzothiophene SERMs, BC-1, could activate ARE signaling pathways via reactive electrophilic metabolite formation. There was no evidence supporting the involvement of the ERs in this process, but the key regulatory factor Nrf2 was demonstrated to be involved in ARE activation by BC-1. These observations suggested a possible correlation between the induction of antioxidant signaling pathways by benzothiophene compounds and their ability to form reactive quinoids; therefore, we have elucidated a novel mechanism of antioxidant signaling activation by benzothiophene compounds via reactive metabolite formation.

Author contribution

Bo-lan YU conceived the study and wrote the manuscript; Zi-xin MAI conducted the cell experiments and analyzed the data; Xu-xiang LIU performed the LC-MS/MS experiments and analyzed the data; Zhao-feng HUANG supervised the entire study and revised the manuscript.