Overexpression of BQ323636.1 Modulated AR/IL-8/CXCR1 Axis to Confer Tamoxifen Resistance in ER-Positive Breast Cancer

NCOR2 is a co-repressor for estrogen receptor (ER) and androgen receptor (AR). Our group previously identified a novel splice variant of NCOR2, BQ323636.1 (BQ), that mediates tamoxifen resistance via interference of NCOR2 repression on ER. Luciferase reporter assay showed BQ overexpression could enhance the transcriptional activity of androgen response element (ARE). We proposed that BQ employs both AR and ER to confer tamoxifen resistance. Through in silico analysis, we identified interleukin-8 (IL-8) as the sole ERE and ARE containing gene responsiveness to ER and AR activation. We confirmed that BQ overexpression enhanced the expression of IL-8 in ER+ve breast cancer cells, and AR inhibition reduced IL-8 expression in the BQ overexpressing cell lines, suggesting that AR was involved in the modulation of IL-8 expression by BQ. Moreover, we demonstrated that IL-8 could activate both AKT and ERK1/2 via CXCR1 to confer tamoxifen resistance. Targeting CXCR1/2 by a small inhibitor repertaxin reversed tamoxifen resistance of BQ overexpressing breast cancer cells in vitro and in vivo. In conclusion, BQ overexpression in ER+ve breast cancer can enhance IL-8 mediated signaling to modulate tamoxifen resistance. Targeting IL-8 signaling is a promising approach to overcome tamoxifen resistance in ER+ve breast cancer.


Introduction
Breast cancer has long been the most prevalent cancer among women. In 2020, around 2.3 million women were diagnosed with breast cancer, accounting for 24.5% of all female cancer cases [1]. In 2020, breast cancer resulted in 685,000 female deaths, contributing to 15.5% of cancer deaths in women (World Health Organization, 2021). The incidence of breast cancer varies among different regions, ranging from 26.2 per 100,000 women in Central South Asia to 95.5 per 100,000 women in Australia/New Zealand [1]. The incidence of breast cancer is increasing on average 0.3% each year over 2009-2018. (National Cancer Institute, US, 2021). Based on gene expression profiles, breast cancer can be classified into five major subtypes, (1) luminal A, (2) luminal B, (3) HER2-enriched, (4) basal-like, and (5) normal-like [2]. The treatment of breast cancer is guided by the expression of estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2). Usually, tamoxifen is used for ER-positive (ER+ve) breast cancer, Herceptin is used for HER2-overexpressed (HER2+ve) breast cancer, and chemotherapy is used for triple-negative breast cancer (TNBC).
About 75% of breast cancer patients belong to ER+ve [3]. ER is the primary tumor driver in this category. Estrogen promotes breast carcinogenesis by binding to ER to stimulate genomic and non-genomic activities essential for cancer-cell proliferation and growth [4]. The genomic pathway refers to ER activation within the cytoplasm to induce in methanol at 1 mg/mL. Recombinant interleukin-8 (IL-8; 208-IL; R&D Systems, Inc., Minneapolis, MN, USA) protein was purchased and dissolved in double-distilled water. Bovine Serum Albumin (BSA; A7030; Sigma-Aldrich, St. Louis, MO, USA) was purchased and dissolved in double-distilled water. Bicalutamide (S1190; Selleckchem, Houston, TX, USA) was dissolved in DMSO at 100 mM as stock concentration.

Cell Viability Assay
MTT assay (M6494; Thermo Fisher Scientific, Waltham, MA, USA) was performed, and 5000 cells were seeded in 96-well plate. A clonogenic assay was performed, and 2000 cells were seeded in 12-well plate. We used 0.01% of crystal violet (C0775; Sigma-Aldrich, St. Louis, MO, USA) to stain the cell colonies. Cells were seeded on 24 wells. The colonies were stained with 0.01% crystal violet and counted under a microscope. A colony with more than 50 cells was regarded as a colony. All experiments were performed in triplicate. Microplate reader Infinite F200 (Tecan, Seestrasse, Switzerland) was used to record the absorbance.

RNA Extraction, Reverse Transcription and qPCR
RNA was isolated using TRIzol (15596026; Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. Briefly, 1 × 10 6 cells were homogenized and lysed in 400 µL TRIzol reagent and mixed with 200 µL chloroform. The aqueous phase that contains the RNA was obtained by centrifugation. RNA was subsequently precipitated with isopropanol and washed with 75% ethanol. RNA was solubilized in DEPC-treated water and the concentration was measured by NanoDrop 1000 (Thermo Fisher Scientific, Waltham, MA, USA). A cDNA synthesis was performed using PrimeScript TM RT Master Mix (RR036B; Takara Biomedical Technology Co., Ltd., China). Briefly, 0.5 µg of RNA was mixed with 5X PrimeScript RT Master Mix and RNase-free water up to 10 µL total volume and incubated at 37 • C for 15 min, and the reaction was heat inactivated at 85 • C for 5 s. The synthesized cDNA was diluted with 40 µL RNase-free water. Samples of cDNA were analyzed by qPCR using the StepOne Real-Time PCR system (Thermo Fisher Scientific, Waltham, MA, USA). Reactions were performed in 10 µL volumes with diluted cDNA, PowerUp™ SYBR™ Green Master Mix (A25742; Thermo Fisher Scientific, Waltham, MA, USA) and qPCR primers. The qPCR primers were shown in the Supplementary Table  (Table S1). The ∆∆CT method was employed to determine the relative expression of a gene.

IL-8 Measurement and AKT Activity Assay
The IL-8 Human ELISA Kit (KHC0081; Thermo Fisher Scientific, Waltham, MA, USA) was employed to determine the amount of interleukin-8 in the culture medium; 50 µL of the culture medium was used. Absorbance at 450 nm was recorded by microplate reader Infinite F200 (Tecan, Seestrasse, Switzerland). AKT Kinase Activity Assay Kit (ab139436; Abcam, Cambridge, UK) was used for determining AKT kinase activity. The signal was developed according to the manufacturer's instructions. Absorbance at 450 nm was recorded by microplate reader Infinite F200 (Tecan, Seestrasse, Switzerland).

Xenograft
Female nude mice at the age of 5 to 6 weeks were used for this study. On the day of inoculation, the cell mixture containing 1 × 10 6 ZR-75-BQ cells was implanted into the mice's abdominal mammary fat pad. The cell mixture was prepared by mixing 50 µL of the cell suspension containing 1 × 10 6 cell with 50 µL of Matrigel (356234; BD Bioscience, Franklin Lakes, NJ, USA), and the 100 µL of the cell mixture was injected into the mammary fat pad. When the tumors were palpable, mice were randomized into 5 groups: (1) Saline (n = 4); (2) 4-OHT (N = 4); (3) 15 mg/Kg repertaxin (n = 4); (4) 4-OHT + 15 mg/Kg repertaxin (n = 4) and; (5) 4-OHT + 15 mg/Kg repertaxin (n = 4). 0.5 mg of tamoxifen dissolved in peanut oil (Sigma-Aldrich, St. Louis, MO, USA) and/or repertaxin (7.5 mg/kg and 15 mg/Kg) by subcutaneous injection twice per week. The tumor sizes were measured regularly using a caliper, and the tumor volume was calculated as the longest diameter x (shortest diameter) 2 /2. At the endpoint of the experiments, mice were euthanized, and tumors were harvested. All the procedures were reviewed and approved by the HKU Committee on the Use of Live Animals in Teaching and Research (CULATR Number: 5140-19).

Immunohistochemistry
Tissue microarray analysis (TMA) was approved by the Institutional Review Board of The University of Hong Kong/Hospital Authority Hong Kong West Cluster (HKU/HA HKW IRB No. UW 08-147). Histological sections were reviewed by the pathologist. For each case, donor blocks were chosen from the representative paraffin tumor blocks, and the selected region was marked for the construction of the TMA block. Clinical data were retrieved from the Department of Pathology, Queen Mary Hospital of Hong Kong. A total of 137 cases (Table S2) were used for scoring of BQ323636.1 and CXCR1 staining. Each case was constructed as triplicates, and the average score was used for the case. TMA sections were deparaffinized by xylene incubation and rehydrated by ethanol. Antigen retrieval was completed by using 0.01 M citrate buffer (pH 6). The slides were put into 3% H 2 O 2 to quench endogenous peroxidase. The slides were rinsed with PBST twice, followed by incubation with primary monoclonal CXCR1 antibody slides were mounted. TMA slides were visualized by the Aperio ScanScope system (Leica Biosystems, Wetzlar, Germany). Two individuals were assigned to finish the scoring. Cytoplasmic expression of CXCR1 was scored, whereas nuclear expression was scored for BQ. The intensity of cytoplasmic CXCR1 was scored as follows: 1 = weak, 2 = moderate, 3 = strong. For the percentage of staining, it was scored as follows: 1 ≤ 25%, 2 ≤ 50%, 3 ≤ 75%, 4 > 75%. The final score was calculated as follows: the score of intensity x the score of percentage. The H-score for nuclear expression was used for BQ323636.1, calculated as follows: 1x% of cells stained at "low" intensity + 2x% of cells stained at "moderate" intensity + 3x% of cells stained at "high" intensity. The median of the H-score was set as the threshold, which was 110 for nuclear BQ and 6.667 for cytoplasmic CXCR1.

In Silico Analysis
Transcriptional Regulatory Element Database (http://rulai.cshl.edu/cgi-bin/TRED/ tred.cgi?process=home; assessed on 18 May 2019) was used to identify genes with androgen response element (ARE) and estrogen response element (ERE), which are androgen receptor targeted-gene and estrogen receptor-targeted gene, respectively. The default setting was used. Lists of genes targeted by AR and ER were retrieved. Genes with both ERE and ARE were identified by comparing the two lists of target genes.

Statistical Analysis
All numerical data were processed in Excel (Microsoft), Prism5 (GraphPad) or SPSS25 (IBM). Data were expressed as mean ± SD from at least three independent experiments. The Mann-Whitney U test or the Students' t-test were performed to compare the variables of the two sample groups. One-way ANOVA and two-way ANOVA were employed to determine the statistical significance for multiple groups. The statistical significance between any two groups was determined by Bonferroni's multiple comparison test. All tests were two-sided unless otherwise specified. Chi-square (χ 2 ) test was used for hypothesis testing. Correlation with survival study of Tissue Microarray expression data was analyzed by Kaplan-Meier estimates followed by the log-rank test carried out by SPSS. Cox proportional hazards regression was used to estimate the association between clinical-pathological parameters, or BQ and CXCR1 scores with survival. Relative risk (RR) and 95% confidence interval (CI) were reported. The proportional-hazards assumption was tested using the Omnibus test, and no major model violation was observed. We considered p < 0.05 considered statistically significant; *, **, and *** represent p < 0.05, p < 0.01 and p < 0.001.

Overexpression of BQ Could Activate AR Signalling and Thus Modulate the Response to Tamoxifen in Breast Cancer
High expression of BQ was found in the tamoxifen-resistant LCC2 cell line ( Figure S1a; Figure S6e), which was consistent with our previous studies [15,23]. BQ interacted with NCOR2 and compromised the repressor function of NCOR2 [15]. Since NCOR2 is a repressor of estrogen receptor α (ER), the presence of excess BQ could suppress its repressive activity, leading to the activation of ligand-independent activation of ER signaling [15]. NCOR2 is also a repressor of androgen receptor (AR) [21]. Therefore, we speculated that overexpression of BQ (Figure 1a; Figure S6a) would induce the activation of AR signaling in breast cancer cells. Through the luciferase reporter assay, we confirmed that BQ overexpression could enhance the transcriptional activity of the androgen response element (ARE) in MCF-7 ( Figure 1b) and ZR-75 cells (Figure 1c). Similarly, we found that ARE activity in LCC2 was significantly higher than that in MCF-7 ( Figure S1b). Next, we determined the effect of BQ down-regulation ( Figure 1d) on ARE activity in LCC2, a tamoxifen-resistant cell line with high expression of endogenous BQ. The results from the reporter assay indicated that knockdown of BQ could reduce ARE activity in LCC ( Figure 1e). These results suggest that BQ can modulate the activity of the AR-driven pathway in breast cancer cells.  Figure 1c). Similarly, we found that ARE activity in LCC2 was significantly higher than that in MCF-7 ( Figure S1b). Next, we determined the effect of BQ down-regulation ( Figure 1d) on ARE activity in LCC2, a tamoxifen-resistant cell line with high expression of endogenous BQ. The results from the reporter assay indicated that knockdown of BQ could reduce ARE activity in LCC ( Figure  1e). These results suggest that BQ can modulate the activity of the AR-driven pathway in breast cancer cells. reporter assay with androgen response element (ARE) was employed. Results were shown as mean ± SD from 6 independent experiments. Student's t-test was employed to determine statistical significance. *** represents p < 0.001. (d) Knockdown efficiency of siRNAs against BQ. LCC2 was transfected with 25 μM of non-targeting siRNA (siCtrl), siBQ.1 or siBQ.2. qPCR was performed 72 h posttransfection. Actin was used as the internal control. Results were shown as mean ± SD from 4 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between 2 groups. ** and *** represent p < 0.01 and p < 0.001 respectively. (e) Knockdown of BQ could reduce AR activity in LCC2. Luciferase reporter assay with ARE was used. Results were shown as mean ± SD from 6 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between 2 groups. *** represents p < 0.001.
Next, we assessed whether AR activity could modulate tamoxifen resistance by comparing the response to tamoxifen in control LCC2 and LCC2 with BQ down-regulation. First, we confirmed that knockdown of BQ within 96 h did not affect cell viability ( Figure  S2). Next, we performed an MTT assay to determine if knockdown of BQ would affect Actin was used as the internal control. Results were shown as mean ± SD from 4 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between 2 groups. ** and *** represent p < 0.01 and p < 0.001 respectively. (e) Knockdown of BQ could reduce AR activity in LCC2. Luciferase reporter assay with ARE was used. Results were shown as mean ± SD from 6 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between 2 groups. *** represents p < 0.001.
Next, we assessed whether AR activity could modulate tamoxifen resistance by comparing the response to tamoxifen in control LCC2 and LCC2 with BQ down-regulation. First, we confirmed that knockdown of BQ within 96 h did not affect cell viability ( Figure S2). Next, we performed an MTT assay to determine if knockdown of BQ would affect tamoxifen response. The results from MTT suggested that knockdown of BQ could make LCC sensitive to tamoxifen (Figure 2a). A Clonogenic assay revealed similar results (Figure 2b). To further consolidate our findings, we employed AR antagonist bicalutamide (BIC). We found that 1 µM of BIC was the maximal non-lethal dosage of bicalutamide in a normal breast cell line MCF-10A as revealed by the MTT assay ( Figure 2c). Next, we confirmed that neither 1 µM of bicalutamide nor 4 µM of tamoxifen affected the cell viability of LCC2 ( Figure 2d). However, co-treatment of different concentrations of bicalutamide and 4 µM of tamoxifen could recover response to tamoxifen in LCC2 in a dose-dependent manner ( Figure 2d). As further confirmation, we treated MCF-7 and ZR-75 with 0.1 nM of DHT to activate AR signaling and determine the response to tamoxifen in these activated cells.
The results from MTT showed that activation of AR made MCF-7 ( Figure 2e) and ZR-75 ( Figure 2f) tolerant to tamoxifen, suggesting that activation of AR signaling can confer tamoxifen resistance in breast cancer. confirmed that neither 1 μM of bicalutamide nor 4 μM of tamoxifen affected the cell viability of LCC2 (Figure 2d). However, co-treatment of different concentrations of bicalutamide and 4 μM of tamoxifen could recover response to tamoxifen in LCC2 in a dosedependent manner ( Figure 2d). As further confirmation, we treated MCF-7 and ZR-75 with 0.1 nM of DHT to activate AR signaling and determine the response to tamoxifen in these activated cells. The results from MTT showed that activation of AR made MCF-7 ( Figure 2e) and ZR-75 (Figure 2f) tolerant to tamoxifen, suggesting that activation of AR signaling can confer tamoxifen resistance in breast cancer.
(f) MTT assay was employed to determine cell viability. Results were shown as mean ± SD from 6 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between 2 groups. *** represent p < 0.001.

Identification of IL-8 as a Candidate to Modulate Tamoxifen Resistance in Breast Cancer
NCOR2 is a repressor for both ER and AR. We hypothesized that BQ overexpression, by compromising the repressor activity of NCOR2, would enhance the activities of both ERE and ARE in breast cancer, conferring resistance to tamoxifen. Through in silico analysis, we found 22 genes that contained both ARE and ERE in their promoter regions ( Figure S3). As confirmation, we treated MCF-7 with either 1 nM of E2 or 1 nM of DHT. We used qPCR to determine the expression of these genes. The results showed that only interleukin-8 (IL-8) was responsive to both E2 and DHT ( Figure S3). Co-treatment of 1 nM of E2 and 1 nM of DHT further enhanced IL-8 expression in MCF-7 ( Figure 3a) and ZR-75 (Figure 3b). Results from ELISA also confirmed enhanced production of IL-8 protein by this treatment (Figure 3c,d). When we treated LCC2 cells with bicalutamide, the expression of IL-8 was reduced, as revealed by qPCR ( Figure 3e) and ELISA (Figure 3f), thus confirming that IL-8 expression in breast cancer can be governed by ARE and ERE activity.
We next examined whether IL-8 could modulate tamoxifen response. First, we confirmed that the expression of IL-8 mRNA ( Figure S4a) and the amount of IL-8 protein ( Figure S4b) were higher in LCC2 when compared to MCF-7. Next, we found that overexpression of BQ in MCF-7 and ZR-75 could enhance the expression of IL-8 on both mRNA ( Figure S4c,d) and protein levels ( Figure S4e,f). The expression of IL-8 was directly correlated with the expression of BQ, suggesting that BQ regulates the expression of IL-8. We employed siRNA to reduce the expression of IL-8 in tamoxifen-resistant cell lines, MCF-7-BQ ( Figure 4a) and ZR-75-BQ (Figure 4b). We found that down-regulation of IL-8 could reverse tamoxifen resistance in these cell lines (Figure 4c,d). When we knocked down BQ expression in LCC2, IL-8 expression was significantly reduced, as revealed by qPCR ( Figure 4e) and ELISA ( Figure 4f). As previously shown, BQ knockdown made LCC2 sensitive to tamoxifen (Figure 2A). These results suggest that IL-8 is one of the downstream pathways of BQ overexpression that modulates resistance to tamoxifen in breast cancer.

IL-8 Activated the AKT-ERK1/2 Axis to Modulate the Response to Tamoxifen
IL-8 (CXCL8) is a pro-inflammatory chemokine that plays an essential role in inflammation and tumor progression [25,26]. IL-8 exerts its effects by binding to the specific G protein-coupled receptors of CXCR1 and CXCR2 [27]. The binding of IL-8 would activate a series of kinases, leading to the activation of the AKT-ERK1/2 signaling cascade [28,29]. Activation of AKT-mediated signaling cascades is associated with the development of tamoxifen resistance [9]. CXCR1 interacts specifically with IL-8, while CXCR2 can bind with different cytokines [30]. We confirmed that IL-8 treatment could activate the AKT-ERK1/2 axis, as revealed by Western blot in MCF-7 and ZR-75 (Figure 5a; Figure S6b). Moreover, IL-8 treatment could enhance AKT kinase activity (Figure 5b,c) and confer tamoxifen resistance (Figure 5d,e). In contrast, tamoxifen resistance was reversed by IL-8 knockdown in LCC2 (Figure 5f). These results demonstrated that IL-8 would be essential for tamoxifen resistance in breast cancer.

Targeting CXCR1/2 Could Reverse Tamoxifen Resistance in Breast Cancer In Vitro and In Vivo
Repertaxin is a non-competitive allosteric inhibitor of CXCR1/2. To examine whether treatment of repertaxin would reverse tamoxifen resistance, we first confirmed that 100 nM of repertaxin was the maximal non-lethal dosage of repertaxin in MCF-10A (Figure 6a) and employed this concentration to test whether this drug would reverse tamoxifen resistance. Clonogenic assay from MCF-7-BQ (Figure 6b), ZR-75-BQ (Figure 6c), and LCC2 ( Figure 6d) confirmed that repertaxin could reverse tamoxifen resistance. Moreover, Western blot demonstrated that repertaxin reduced the levels of p-AKT and p-ERK1/2 in BQ overexpressing cell lines (Figure 6e; Figure S6c). Actin was used as the internal control. Results were shown as mean ± SD from 6 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between DMSO and any of the treatment groups. (c,d) ELISA was performed to confirm the effect of 1 nM of E2 and 0.1 nM of DHT on the production of IL-8. Culture medium was collected after 24 h of the treatment. ELISA was performed to determine the amount of IL-8 in the medium. Results were shown as mean ± SD from 4 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between untreated and any of the treatment groups. Suppression of AR could reduce the (e) mRNA expression and (f) protein production of IL-8 in LCC2. The cells were treated with 1 µM of bicalutamide (BIC; AR antagonist) for 48 h. qPCR was performed to determine mRNA. Results were shown as mean ± SD from 4 independent experiments. Student's t-test was employed to determine statistical significance. ELISA was performed to evaluate the production of IL-8 in the culture medium. Results were shown as mean ± SD from 4 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between untreated and BIC treated groups. *, **, and *** represent p < 0.05, p < 0.01 and p < 0.001. test was employed to determine statistical significance. ELISA was performed to evaluate the production of IL-8 in the culture medium. Results were shown as mean ± SD from 4 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between untreated and BIC treated groups. *, **, and *** represent p < 0.05, p < 0.01 and p < 0.001.  MTT assay was performed to determine cell viability after 72 h of TAM treatment. Results were shown as mean ± SD from 5 independent experiments. Student's t-test was employed to determine statistical significance between siCtrl and siIL-8 treated groups. Knockdown of BQ could reduce the (e) mRNA expression and (f) protein expression of IL-8 in LCC2. LCC2 cells were treated with the siRNAs. qPCR was performed to determine the mRNA level of IL-8, 48 h post-transfection. Results were shown as mean ± SD from 6 independent experiments. ELISA was performed to determine the amount of IL-8 in the culture medium. Results were shown as mean ± SD from 4 independent experiments. One-way ANOVA was employed. Bonferroni's multiple comparison test was employed to determine the significance between untreated and siRNAs treated groups. *, **, and *** represent p < 0.05, p < 0.01 and p < 0.001. moxifen resistance [9]. CXCR1 interacts specifically with IL-8, while CXCR2 can bind w different cytokines [30]. We confirmed that IL-8 treatment could activate the AKT-ERK axis, as revealed by Western blot in MCF-7 and ZR-75 (Figure 5a; Figure S6b). Moreov IL-8 treatment could enhance AKT kinase activity (Figure 5b,c) and confer tamoxifen sistance (Figure 5d,e). In contrast, tamoxifen resistance was reversed by IL-8 knockdo in LCC2 (Figure 5f). These results demonstrated that IL-8 would be essential for tamoxi resistance in breast cancer.

Targeting CXCR1/2 Could Reverse Tamoxifen Resistance in Breast Cancer In Vitro and I Vivo
Repertaxin is a non-competitive allosteric inhibitor of CXCR1/2. To examine whet treatment of repertaxin would reverse tamoxifen resistance, we first confirmed that nM of repertaxin was the maximal non-lethal dosage of repertaxin in MCF-10A (Fig  6a) and employed this concentration to test whether this drug would reverse tamox resistance. Clonogenic assay from MCF-7-BQ (Figure 6b), ZR-75-BQ (Figure 6c), and LC To further confirm the effect, an in vivo model was employed. Xenografts were established using ZR-75-BQ cells implanted onto the mammary fat pad of female nude mice. The mice were treated with tamoxifen (0.5 mg/mouse; twice/week) and given subcutaneous injection of repertaxin (7.5 mg/Kg and 15 mg/Kg; twice/week; Figure 7a). Compared with the saline control, mice treated with either 4-OHT or repertaxin alone did not show significant tumor reduction (Figure 7b). Mice treated with repertaxin and 4-OHT together could significantly suppress tumor growth, with such an effect being dose-dependent as revealed by two-way ANOVA (Figure 7b; F = 8.58; p < 0.001). Western blot analysis confirmed that treatment of repertaxin could reduce the levels of both p-AKT and p-ERK1/2 in the tumors (Figure 7c; Figure S6d). Our results, therefore, suggest that targeting CXCR1 by repertaxin could recover tamoxifen sensitivity.

Clinical Significance of CXCR1 in Breast Cancer
Immunohistochemistry was employed to assess the expression of CXCR1 and BQ in the primary ER+ve breast tumor (Figure 8a). Cytoplasmic CXCR1 expression was positively correlated with nuclear BQ expression (Figure 8b) shown by the chi-square test, p = 0.029. High cytoplasmic CXCR1 expression was also associated with tamoxifen resistance, relapse, and metastasis (Figure 8c-e). The Kaplan-Meir log-rank text showed it was significantly associated with poor outcome for overall survival (log-rank test; p = 0.006; Figure 8f) and disease-specific survival (log-rank test; p = 0.003; Figure 8g). Univariate Cox regression analysis for overall survival (Table 1) showed that cases with high cytoplasmic CXCR1 were significantly associated with poorer overall survival (RR = 3.171, 95% CI 1.322, 7.610; p = 0.010), but this failed to remain significant on multivariate analysis. Similar findings were obtained for combined analysis of high cytoplasmic CXCR1 and high nuclear BQ expression. Interestingly, a Cox regression analysis for disease-specific survival ( Table 2) showed cases with high cytoplasmic CXCR1 were significantly associated with poorer disease-specific survival both for univariate (RR = 5.350, 95% CI 1.519, 18.840; p = 0.009) and multivariate analyses (RR = 4.661, 95% CI 1.313, 16.545; p = 0.017). Similar findings were also obtained for a combined analysis of high cytoplasmic CXCR1 and high nuclear BQ, with significant association with poorer disease-specific survival in both univariate (RR = 5.401, 95% CI 1.500, 19.449; p = 0.010) and multivariate analyses (RR = 4.860, 95% CI 1.318, 17.919; p = 0.018). These results confirm that cytoplasmic CXCR1 expression could be an independent prognostic factor for disease-specific survival in ER+ve breast cancer.

Discussion
Most breast cancer patients are ER+ve and receive tamoxifen as adjuvant endocrine treatment. Despite the high efficacy of tamoxifen, one-third of these patients still relapse after tamoxifen treatment. In this study, we found that BQ, in disrupting the gene repressor function of NCOR2, could induce the expression of IL-8. We confirmed the presence of functional ERE and ARE in the IL-8 promotor. Excess BQ could compete with the repressor function of NCOR2, reducing its suppressive effect on ER and AR. Activation of both ERE and ARE in the IL-8 promotor intensifies BQ's effect on IL-8 production. Targeting the IL-8 mediated signaling cascade could reverse tamoxifen resistance in vitro and in vivo, illustrating one more possible way to combat tamoxifen resistance in breast cancer.
Our demonstration that BQ overexpression can promote ARE and ERE activities in ER+ve breast cancer cell lines is novel. BQ overexpression can thus enhance both AR and ER-mediated signaling activities that can lead to tamoxifen resistance. Through in silico analysis, we identified 22 candidate genes that contained both ARE and ERE within the promoter region and confirmed that interleukin-8 (IL-8) had both functional ARE and ERE activity ( Figure S1), suggesting that IL-8 could be regulated by both AR and ER in ER+ve breast cancer. Having shown that BQ overexpression can enhance both ERE activity [14] and ARE activity (Figure 1a-e), we next confirmed that stimulating ERE and ARE activity can enhance the expression of IL-8 in ER+ve breast cancer cell lines (Figure 3a-d). Inhibition of AR reduced IL-8 expression in the BQ overexpressing cell lines (Figure 3e-f). These results suggest that the modulation of IL-8 expression by BQ involved AR. We confirmed that IL-8 could activate both AKT and ERK1/2 via CXCR1 to confer tamoxifen resistance; however, we cannot exclude the effect of CXCR2 as repertaxin targets both CXCR1 and CXCR2. The small inhibitor repertaxin could reverse tamoxifen resistance in vitro and in vivo. Our study illustrates one more possible way to combat tamoxifen resistance in breast cancer.
IL-8 is a pro-inflammatory chemokine that has been suggested to promote tumor progression, angiogenesis, and metastasis in cancer [31]. Overexpression of IL-8 is associated with drug resistance. Inhibition of IL-8 has been suggested to reverse paclitaxel and doxorubicin resistance in breast cancer cell lines [32]. It has been reported that the expression of IL-8 was significantly increased in tamoxifen-resistant MCF-7 and ZR-75 cell lines [33]. IL-8 interacts with its receptor CXCR1 to activate downstream signaling pathways [34], such as the AKT [35] and ERK1/2 [36]. Activation of PI3K/AKT has been suggested to induce tamoxifen resistance in ER+ve breast cancer [37], and increased activity of the ERK1/2 pathway has been shown to involve tamoxifen resistance [38]. Furthermore, IL-8 has been demonstrated to activate STAT3 signaling in prostate cancer for promoting the disease progression [39]. Our previous study demonstrated that overexpression of BQ could enhance STAT3 signaling by up-regulating the expression of IL-6 to modulate tamoxifen resistance in breast cancer [22]. Overexpression of BQ might mediate IL-6 and IL-8 signaling; in addition, their crosstalk might contribute to the drug resistance. Our study is the first to describe AR mediating tamoxifen resistance through enhancing the expression of IL-8, which subsequently activates the AKT and ERK1/2 signaling cascade. Furthermore, we showed that repertaxin could compromise AKT and ERK1/2 activities in vitro ( Figure 6) and in vivo (Figure 7). Apart from CXCR1, CXCR2 has been shown to modulate the development, progression and drug response of breast cancer [40]. CXCR2 can also modulate drug resistance. Activation of CXCR2 by IL6 might induce resistance to paclitaxel, while activation of CXCR2 by IL8 might induce resistance to doxorubicin [32]. In our study, we found that repertaxin could reduce tamoxifen resistance. Hence, CXCR1/2 can be a possible target for developing a therapeutic agent against tamoxifen resistance in breast cancer.
For ER+ve breast cancer, ER signaling plays a critical role in disease progression. Activation of ER by its ligand estrogen can trigger transcription of target genes that maintain estrogen response element (ERE), leading to cancer cell growth and proliferation. Therefore, targeting ER is an efficient approach to inhibit ER+ breast cancer. In this study, we revealed that AR could also contribute to tamoxifen resistance. There are 70-90% of breast cancer patients who express the AR [41], with several studies indicating that AR might be a predictive or prognostic factor and a drug target in breast cancer [42]. However, it is still controversial whether AR is a good or bad prognostic factor. AR has been shown to correlate with favorable outcomes, such as smaller tumor size, lower tumor grade, less necrosis, lower Ki-67 levels, and better treatment response in ER+ve breast cancer [43][44][45]. However, the AR-to-ER expression ratio has been considered to affect the prognosis and response to tamoxifen treatment. If the ratio of AR to ER was greater than 2, it had an increased risk of failure with tamoxifen therapy [46]. These findings suggest the role of AR could be context-dependent and should not be regarded as an independent factor contributing to tamoxifen resistance per se. Our study suggests that AR might contribute to intensifying signaling components that modulate tamoxifen resistance. Although the role of AR remains controversial, it is nevertheless an attractive candidate for developing novel breast cancer therapies, potentially improving breast cancer patients' survival outcomes.

Conclusions
In summary, we demonstrated a possible mechanism mediated by BQ to induce tamoxifen resistance in ER+ve breast cancer through AR-mediated signaling. BQ could up-regulate IL-8 to modulate tamoxifen resistance and confer tamoxifen resistance through IL-8 mediated AKT and ERK1/2 pathways. Targeting IL-8 and its receptors would be a potential therapeutic approach to combat tamoxifen resistance in ER+ve breast cancer.
Supplementary Materials: The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/life12010093/s1, Figure S1. Endogenous BQ expression level and basal ARE activity in MCF-7 and LCC2, Figure S2. The effect of BQ knockdown in LCC2, Figure S3. Relative mRNA expression of genes contained both ARE and ERE in their promoter region, Figure S4. Expression of IL-8 in MCF-7 and LCC2 and the effect of BQ overexpression in MCF-7 and ZR-75, Figure S5. Protein quantification of Western blot in this study, Figure S6. Original blots of Western blot of this study. Table S1: List of qPCR primer sequences used in this study, Table S2: Clinical characteristics of breast cancer patients.  Informed Consent Statement: Patient consent was waived as archival tissue was used and all patients' samples were dis-identified for study. Data Availability Statement: Transcriptional Regulatory Element Database (http://rulai.cshl.edu/ cgi-bin/TRED/tred.cgi?process=home; assessed on 18 May 2019) was used to identify genes with androgen response element (ARE) and estrogen response element (ERE). The datasets and materials generated in this study are available from the corresponding author on reasonable request.