ERβ1 Expression Patterns Have Different Effects On EGFR TKIs Treatment Response in EGFR Mutant Lung Adenocarcinomas.

Background: Estrogen receptor β (ERβ) can regulate cellular signaling through non-genomic mechanisms, potentially promoting resistance to epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs). However, the mechanisms underlying the ERβ-mediated resistance to EGFR TKIs remain poorly understood. Methods: qRT-PCR was performed to investigate ERβ1 and ERβ5 expression levels in cell lines. The localization of ERβ and ERβ1 within cells was assessed using immunocytochemistry and immunouorescence. The effect of estradiol and/or getinib on EGFR signaling pathways was determined by western blot. Cell viability and colony formation assays were used to assess getinib response for different cell lines. The apoptosis was veried by tunel and western blot. Immunohistochemistry was used to assess the expression of ERβ1 in lung adenocarcinoma tissues. Patient survival was estimated using the Kaplan-Meier method, and comparisons between groups were conducted using log-rank tests. Results: PC9 cell lines stably overexpressing ERβ1 or ERβ1/ERβ5 were established successfully. Immunouorescence revealed that ERβ5 overexpression partly retained ERβ1 in the cytoplasm. Immunoblotting analyses revealed that EGFR pathway activation levels were higher in PC9/ERβ1/5 cells than those in PC9/ERβ1 or control PC9 cells. In the presence of estradiol, PI3K/AKT/mTOR pathway activation levels were higher in ERβ1/5-expressing cells than those in ERβ1-expressing cells. Additionally, PC9/ERβ1/5 cells were less prone to the cytotoxic and pro-apoptotic effects of getinib compared with PC9/ERβ1 or control PC9 cells. Conclusion: Cytoplasmic ERβ1 was associated with poor progression-free survival in lung cancer patients treated with EGFR TKIs. These results suggest that anti-estrogen therapy might reverse EGFR TKI treatment resistance to some extent in selected patients. under

Preclinical studies have shown that EGFR expression was downregulated in response to estradiol (E2); in contrast, ERβ antagonists upregulated EGFR expression, highlighting the crosstalk between ERβ and EGFR signaling [12]. Hence, it is believed that non-genomic signaling events may modulate EGFR TKI resistance. ER belongs to the nuclear receptor superfamily of ligand-activated transcription factors. Since a nuclear localization of ERβ in cancer cells has been reported [13], the relevance of cytoplasmic ERβ in non-genomic signaling activation in cancer cells has attracted increasing attention in recent years.
Studies on endocrine-related cancers suggested that certain ERβ isoforms are associated with ERβ protein localization and patient prognosis [14][15][16][17]. For example, ERβ1 (also known as wild-type ERβ) was primarily found in the nucleus of prostate cancer cells, whereas ERβ5 localized both in the cytoplasm and nucleus [16,18,19]. Although ERβ5 lacks the ability to bind estrogen or form homodimers due to the absence of helix 12 in its C-terminal, it can heterodimerize with ERβ1 in the presence of estrogen [20].
The aim of this study was to assess the role of the interaction between ERβ1 and ERβ5 in non-genomic signaling in lung adenocarcinoma. To this end, we overexpressed ERβ1 and ERβ5 in EGFR exon 19 deletion-harboring lung adenocarcinoma cells and assessed their ability to form heterodimers, as well as the relevance of ERβ1/ERβ5 heterodimerization in non-genomic signaling and response to EGFR TKIs.

Materials And Methods
Cell culture and chemicals The EGFR-mutant lung adenocarcinoma cells PC9, HCC827, H1975, and H1650 were kindly provided by Peking University Cancer Hospital. Cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum and maintained at 37°C in a humidi ed 5% CO 2 atmosphere.
Ge tinib was purchased from Selleck Chemicals (Selleck, USA) and diluted in dimethyl sulfoxide (DMSO) at a concentration of 10 mmol/L. Estradiol (E2) was purchased from Sigma-Aldrich (Sigma-Aldrich, Germany) and diluted in pure ethanol at a concentration of 10 mmol/L. Both drugs were aliquoted and stored at -80°C.
Establishment of cell lines stably expressing ERβ1 and ERβ1/5 Lentiviral vectors expressing ERβ1 and ERβ5 were purchased from GenePharma (Shanghai, China). PC9 cells were infected with lentiviruses (MOI = 50) for three days. Subsequently, transduced cells were selected with 2 μg/mL of puromycin for one week. ERβ1-overexpressing single-cell clones were established (hereafter referred to as PC9/ERβ1 cells), and stable ERβ1 overexpression was con rmed by Western blot and qRT-PCR. PC9/ERβ1 cells were then infected with viruses carrying ERβ5 open reading frame (ORF), followed by selection with neomycin (600 mg/mL) for one week.
Cell viability and colony formation assays Cells were treated with estradiol (20 nM) during the experiment. Cell viability was assessed using a cell counting kit-8 (CCK8; Dojindo, Japan). Brie y, cells were seeded (3×10 3 cells/ well) in sextuplicate in 96well plates containing 100 μL medium and incubated for 24 hours. Subsequently, cells were treated with increasing concentrations of the indicated drugs for an additional 72 hours. After treatment, 10 μL of water-soluble tetrazolium salt (WST-8) was added to each well and incubated for 2 hours. Optical absorbance at 450 nm was measured using a microplate reader. Relative viability was calculated using the following formula: Relative viability (%/control) = [A450 (treated)-A450 (blank)]/[A450 (control) -A450 (blank)].
For colony formation assays, cells were seeded into 6 cm cell culture dishes (500 cells/dish) and treated for two weeks with 40 nM ge tinib or DMSO (1/1000 dilution). After washing twice with phosphatebuffered saline (PBS), cells were stained with crystal violet (Beyotime, China) for 20 min and washed with PBS.
Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) Cells were seeded on glass slides in 6-well plates 3×10 4 cells/well. After a 24 hour treatment with 40nM ge tinib or DMSO, cells were xed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate. Then, 50 μL of a freshly prepared TUNEL solution (Keygene, China) was added onto each slide. Subsequently, cell nuclei were counterstained with DAPI, and samples were imaged using a uorescence microscope (Zeiss, Germany). The percentage of apoptotic cells normalized to the control group (set to 100%) was calculated after counting cells in ve representative elds. The data were expressed as mean ± standard deviation (SD).

Patients
The data from 103 Chinese patients with advanced lung adenocarcinoma were retrospectively reviewed. The inclusion criteria used for patient enrollment were as follows: (1) Pathological diagnosis of adenocarcinoma; (2) su cient tissue for both EGFR and KRAS mutation detection and ERβ1 immunohistochemistry; (3) presence of EGFR mutations associated with sensitivity to EGFR TKIs, including 19 exon deletion and 21 exon point mutation, and absence of EGFR T790M or KRAS mutations; (4) patients treated with EGFR TKIs, including erlotinib, ge tinib, and icotinib; (5) available clinicopathological characteristics, including sex, age, disease stage, and smoking history. Treatment responses were classi ed according to the response evaluation criteria in solid tumors (RECIST), version 1.1. Progression-free survival (PFS) time was de ned as the time between the rst day of EGFR TKI treatment until radiologic progression or death. The study was approved by the Ethics Review Committee of the Shandong Cancer Hospital.

EGFR and KRAS mutation detection and immunohistochemistry for ERβ1
Ampli cation refractory mutation system (ARMS) was employed to detect different genetic variants, including EGFR (exon 19 deletions, L858R, and T790M) and KRAS mutations.
ERβ1 expression in lung adenocarcinoma tissue samples was assessed by immunohistochemistry (IHC).
Informed consent to use biopsy tissues was obtained from all patients. Brie y, formalin-xed, para nembedded tissue sections (3-μm) were depara nized and stained according to standard procedures.
Sections were probed with anti-ERβ1 mouse antibody (1:200; Abcam, USA); a biotinylated anti-mouse IgG secondary antibody was used. Brown staining in cytoplasm or/and nucleus was considered positive. No staining was observed in negative controls, including lung tissues probed with a non-immune primary antibody. Based on the localization of "positive" immunoreactivity in the cytoplasm, nucleus, or both, patients were grouped as cERβ1-, n/cERβ1-, or nERβ1-positive. IHC staining was evaluated independently by two investigators (Lijuan Zhang and Meng Tian) and a pathologist (Jianbo Zhang).

Statistical analysis
Differences in the relative mRNA levels, cell viability, and apoptosis between different cell lines were analyzed using two-tailed Student's t-tests. Patient survival was estimated using the Kaplan-Meier method, and comparisons between groups were conducted using log-rank tests. All statistical tests were two-tailed, and P-values < 0.05 were considered statistically signi cant. All statistical analyses were performed using GraphPad Prism 8.0. (Prism Software Inc., San Diego, USA).

ERβ5 affects ERβ1 localization in EGFR-mutant lung adenocarcinoma cancer cells
ERβ5 has been identi ed as the predominant ERβ splice variant in non-malignant lung tissues. In this study, we found that ERβ5 mRNA levels were elevated in four lung adenocarcinoma cell lines harboring EGFR mutations ( Figure 1A). We further assessed the role of ERβ5 in lung adenocarcinoma using PC9 cells, which harbor EGFR exon 19 deletions. Immunocytochemistry and immuno uorescence analyses revealed that endogenous ERβ predominantly localized in the cell cytoplasm, and only low ERβ levels were detected in the nucleus ( Figure 1B, 1C).
Next, we overexpressed ERβ1 in PC9 cells (PC9/ERβ1); we also overexpressed ERβ5 in PC9/ERβ1 cells (hereafter referred to as PC9/ERβ1/5). ERβ1 and ERβ5 overexpression were con rmed at the mRNA and protein levels by qRT-PCR and immunoblotting, respectively ( Figure 1D; Table 1). Immuno uorescence using a non-variant speci c antibody revealed that ERβ levels were elevated both in PC9/ERβ1 and PC9/ERβ1/5 cells; however, ERβ localization differed between the two cell lines. Although ERβ primarily localized in the cell nucleus in PC9/ERβ1 cells, in PC9/ERβ1/5 cells, it was found both in the cytoplasm and nucleus (Figure 2A).
ERβ1 has the highest a nity for estradiol among all ERβ splice variants. Hence, we used an ERβ1-speci c antibody to determine ERβ1 localization. ERβ1 predominantly localized in the cell nucleus in PC9/ERβ1 cells. However, in PC9/ERβ1/5 cells, we observed that ERβ1 was partly detained in the cytoplasm, suggesting that the expression of ERβ5 suppressed ERβ1 translocation from the cytoplasm to the nucleus ( Figure 2B).
The interaction between ERβ1 and ERβ5 regulates downstream signaling events in the presence of estradiol Next, we assessed the role of nuclear and cytoplasmic ERβ in transcriptional regulation and non-genomic signaling, respectively. The expression of the cell cycle regulator P21 is induced by the nuclear ERβ [18].
In this study, we found that P21 expression levels were profoundly higher in PC9/ERβ1 cells compared to those in PC9/NC or PC9/ERβ1/5 cells, especially after stimulation with estradiol ( Figure 3A). PI3K/AKT/mTOR signaling pathway is regulated by both EGFR and ERβ [11]. To determine the PI3K/AKT/mTOR pathway activation status, we assessed the levels of EGFR, AKT, and RPS6. We found that phospho-EGFR levels were lower in PC9/ERβ1/5 cells than those in PC9/NC or PC9/ ERβ1 cells. Although total and phospho-EGFR levels decreased in all groups after estradiol treatment, the decrease in phospho-EGFR levels was stronger in PC9/ERβ1 and PC9/ERβ1/5 cells than that in PC9/NC cells. The phospho-AKT levels were higher in PC9/ERβ1/5 cells than those in PC9 or PC9/ERβ1 cells, both at baseline and after estradiol treatment. Although ERβ1 overexpression had a limited impact on baseline phospho-AKT levels, the increase in phospho-AKT levels after estradiol treatment was stronger in ERβ1overexpressing cells compared with that in PC9/NC cells. The levels of phospho-RPS6, which functions downstream of mTOR, were similar among the groups ( Figure 3A).
We also found that ERβ1 but not ERβ5 was upregulated in PC9/NC cells after estradiol treatment. Interestingly, qRT-PCR showed no changes in the ERβ1 mRNA levels after estradiol treatment, suggesting that the estradiol-mediated ERβ1 upregulation occurs at the post-transcriptional level ( Figure 3B).
When cells were treated with ge tinib in addition to estradiol, phospho-EGFR levels were decreased in all groups, whereas phosphor-AKT levels were increased, especially in PC9/ERβ1/5 cells. Similar to phospho-EGFR, phospho-RPS6 levels were decreased in all groups after ge tinib treatment. P21 was also downregulated in ge tinib-treated cells ( Figure 3C).

PC9/ERβ1/5 cells are less prone to the cytotoxic effects of ge tinib
To determine the effects of different ERβ splice variants on response to ge tinib, we performed cell viability and colony formation assays. We found that PC9/ERβ1/5 cells were less prone to the cytotoxic effects of ge tinib (40 nM) compared with PC9/NC or PC9/ERβ1 cells, although we found no signi cant differences in cell viability at low concentrations of ge tinib ( Figure 4A). Additionally, the ability of ge tinib (40 nM) to inhibit colony formation was stronger in PC9/ERβ1 and PC9/NC than in PC9/ERβ1/5 cells ( Figure 4B).
Ge tinib treatment induced apoptosis in all three groups. However, the pro-apoptotic effects of ge tinib were more potent in PC9/ERβ1 and PC9/NC cells compared with those in PC9/ERβ1/5 cells ( Figure 4C). We also measured cleaved-PARP and cyclin D3 levels, which are commonly used as markers of apoptosis. We found that cPARP levels were increased in all three groups after ge tinib treatment. However, the increase in cPARP levels was more substantial in PC9/NC and PC9/ERβ1 cells than in PC9/ERβ1/5 cells. Consistently, the decrease in cyclin D3 levels was more profound in PC9/NC and PC9/ERβ1 cells, while almost no change in cyclin D3 levels was observed in PC9/ERβ1/5 cells after ge tinib treatment ( Figure 4D). These results suggest that PC9/ERβ1/5 cells are less sensitive to EGFR TKIs than PC9/NC and PC9/ERβ1 cells.

ERβ1 expression and intracellular distribution affect PFS in patients with advanced EGFR-mutant lung adenocarcinoma
In this study, we retrospectively analyzed the data from 103 stage IIIb-IV lung adenocarcinoma patients treated with EGFR TKIs at the Shandong Cancer Hospital between January 2014 and November 2017. All patients harbored EGFR mutations affecting response to EGFR TKIs, including exon 19 deletions (47; 45.6%) and exon 21 point mutations (55; 53.4%); one patient had G719X mutations, whereas no EGFR T790M or KRAS mutations were detected. The clinicopathological characteristics of the patients are summarized in Table 2. Most patients were never/light smokers (79; 76.7%) and women (65; 63.1%).
Because cytoplasmic ERβ1 is key for non-genomic signaling activation, we combined patients with n/cERβ1 expression and those with cERβ1 (cytoERβ1). Survival analysis showed that patients with cytoERβ1 expression (n = 43) had a shorter median PFS after EGFR TKI treatment (9.5 months) compared to those with nERβ1 expression (n =37; PFS, 17.5months; P = 0.0003) ( Figure 4E).

Discussion
Approximately 20%-30% of patients with EGFR activating mutations exhibit primary resistance to EGFR TKIs. The mechanism underlying resistance to EGFR TKIs, and primary resistance, in particular, are extremely complex and remain poorly understood. ERβ expression has been associated with response to EGFR TKIs. Notably, in a Japanese cohort study, strong ERβ expression predicted favorable clinical outcomes in patients with lung adenocarcinoma after treatment with EGFR TKIs. In contrast, we previously identi ed high cytoplasmic ERβ expression as a predictor of poor PFS [21,22]. Therefore, further elucidation of the expression pattern and intracellular distribution of ERβ is required to determine the effects of non-genomic signaling on EGFR signal transduction and clinical outcomes.
Several ERβ splicing variants have been identi ed, the most important of which are ERβ1 (wild-type ERβ), and ERβ2-5 [20,23]. ERβ1 is the only fully functional receptor in the ERβ family, and has the highest a nity for estradiol; other ERβ family members have weak to no ligand binding capacity, despite maintaining their ability to heterodimerize with ERβ1 [20]. Therefore, assessing the function of ERβ splice variants other than ERβ1 is equally important. Notably, the crucial role of ERβ5 in lung cancer is becoming increasingly evident [17,24].
In this study, we focused on the role of ERβ1 and ERβ5 in lung adenocarcinoma. Previous studies demonstrated that ERβ1 was predominantly localized in the cell nucleus and exerted anti-proliferative effects. In contrast, ERβ5 was found both in the cytoplasm and nucleus, and it has been implicated in cancer cell migration and invasion [17,22,25]. Our results con rmed the elevated ERβ5 levels in EGFRmutant lung cancer cells; in contrast, ERβ1 was lowly expressed. These results were consistent with those of a previous study showing that ERβ5 was the primary ERβ isoform expressed in non-malignant lung cells, and heterodimerized with ERβ1 [20]. Similarly, we previously showed that ERβ5 formed complexes with ERβ1, con rming their ability to interact [22].
In this study, we also found that ERβ1 was predominantly localized in the cell nucleus. However, the forced overexpression of ERβ5 partly retained ERβ1 in the cytoplasm. Hence, the presence of ERβ5 can explain previous ndings of ERβ1 localization in the nucleus and cytoplasm in cancer cells.
Total and phospho-EGFR levels were decreased after estradiol treatment, highlighting the crosstalk between EGFR and ERβ signaling pathways [12]. P21 is an essential cell cycle regulator, playing important tumor-suppressing roles [26]. Importantly, P21 expression was induced by ERβ [18,25]. In this study, we found that con rmed that ERβ1 increased P21 levels, suggesting a role of ERβ1 in transcriptional regulation in lung cancer cells. Consistently, ERβ1 exerted anti-proliferative effects in other cancer cells [15,18]. However, when ERβ1 and ERβ5 were co-expressed, P21 levels were lower compared with those in PC9/ERβ1 cells, suggesting that ERβ5 impairs the transcriptional abilities of ERβ1. However, in the presence of estradiol, PI3K/AKT/mTOR signaling pathway activation levels were higher in ERβ1/5expressing lung cancer cells than those in ERβ1-expressing cells, suggesting that the interaction between ERβ1 and ERβ5 potentiated the effects of ERβ1 in non-genomic signaling. Hence, we believe that ERβ1 translocation from the nucleus to cytoplasm in the presence of ERβ5 was essential in determining its biological function, re ecting the bi-faceted role of ERβ in cancer [27]. mTOR signaling was inhibited by ge tinib treatment in all groups, although phospho-AKT levels were increased. Consistently, ge tinib treatment exerted cytotoxic effects in all cell groups. However, ERβ1/ERβ5 co-expression rendered cells less prone to the cytotoxic and pro-apoptotic effects of ge tinib. These results con rmed the critical role of ERβ1/ERβ5 complexes in estrogen receptor-mediated nongenomic signaling. We also found that estradiol upregulated ERβ1 but not ERβ5 at the posttranscriptional level, con rmed the high a nity of ERβ1 for estradiol.
In this study, we also investigated the effect of the ERβ1 expression pattern on PFS in EGFR-mutant lung adenocarcinoma patients. We found that patients with nuclear ERβ1 expression exhibited a relatively longer PFS after EGFR TKI treatment, whereas cytoplasmic ERβ1 was associated with shorter PFS after EGFR TKI treatment. These results highlight the clinical relevance of our ndings from in vitro experiments in EGFR-mutant lung cancer cell lines. Importantly, nuclear ERβ1 expression in lung cancer tissues was associated with tumor-suppressing effects, whereas cytoplasmic ERβ1 promoted EGFR TKI resistance to some extent. Although cytoERβ1 was associated with a lower response to EGFR TKIs, the median PFS of patients with cytoERβ1 was 9.5 months, suggesting that EGFR mutations remain the most powerful predictor for EGFR TKI treatment response. The ndings reported here need to be con rmed in large cohort prospective studies. Additionally, the relationship between ERβ1/ERβ5 ratio and response to EGFR TKIs merits further investigation.

Conclusion
We showed that ERβ1 localized in the cell cytoplasm by interacting with ERβ5, inducing non-genomic signaling activation, and promoting EGFR TKI treatment resistance in EGFR-mutant lung adenocarcinoma. Hence, anti-estrogen therapy might reverse EGFR TKI treatment resistance to some extent in certain patients.