ERβ Regulates NSCLC Phenotypes by Controlling Oncogenic RAS Signaling

This article is featured in Highlights of This Issue, p. 813

Lung cancer is the leading cause of cancer-related death in both men and women worldwide. Although the major risk factor is tobacco smoking, there is a significant proportion of never smokers that develop lung cancer (1). Observations from population-based clinical studies propose a role for female steroid hormones in lung tumor development and progression. The nonsmoking related lung cancer is more common in women, and premenopausal women develop less differentiated lung cancer compared with postmenopausal women that have lower levels of circulating estrogen (2, 3). Interestingly, local production of estradiol has been observed in non–small cell lung cancer (NSCLC). Its concentration is higher in cancer tissues compared with nonneoplastic lung tissues and its intratumoral concentration has been associated positively with aromatase expression and markers of tumor growth in a group of male and postmenopausal female patients with NSCLC (4). Consistent with the clinical studies, treatment with estrogen was reported to promote progression of p53-defective mouse lung tumors that express mutant k-ras (5).

Estrogens regulate a variety of physiologic processes, including cell growth, differentiation, and development. Estrogens mediate their actions through two members of the nuclear receptor superfamily, estrogen receptor (ER) α and β. In response to ligand binding or in a ligand-independent manner ERs can regulate gene expression either by acting as transcription factors at sequence-specific response elements known as estrogen response elements or by interacting with and activating other transcription factors (6). ERs demonstrate different tissue distribution and perturbation of ER subtype-specific expression has been detected in various pathologic conditions, including cancer. Whereas ERα is overexpressed in a significant proportion of breast cancers, both ERs have been detected in NSCLC cells (6). However, the role of ERs in NSCLC remains poorly understood because previous studies produced contradictory data. Although two cell-based studies have reported an increased NSCLC cell proliferation in response to treatment with ERβ ligands, clinical studies demonstrated a correlation between ERβ positivity and better outcome of patients with lung cancer (7–12). In particular, increased expression of wild-type ERβ (ERβ1) has been associated with better prognosis and reduced mortality and is inversely associated with lymph node metastases and tumor size in patients with NSCLC (11, 13, 14). In addition, the correlation of ERβ1 with increased disease-free survival in patients with NSCLC carrying EGFR mutations and better response to EGFR tyrosine kinase inhibitors (EGFR–TKI) proposed an antitumorigenic function of ERβ in NSCLC that involves potential regulation of growth factor signaling (9, 10).

EGFR that is expressed in high levels in 62% of NSCLCs correlates with poor prognosis (15). Furthermore, NSCLCs often produce EGFR ligands such as EGF and TGF-α, suggesting the function of an autocrine growth-stimulatory mechanism that supports the EGFR oncogenic actions (16). Antibody-based therapies that target the EGFR ligand–binding domain and disrupt the autocrine receptor-activating mechanisms have been associated with improved survival in patients with lung cancer (17). In addition to receptor overexpression, somatic mutations in the tyrosine kinase domain that result in constitutive activation of EGFR were identified in NSCLCs. The EGFR–TKIs gefitinib and erlotinib that bind preferentially to the tyrosine kinase domain of some of the EGFR mutants and inhibit the activity of the receptor have been associated with improved clinical outcome in patients with EGFR-mutant lung tumors (18). Unfortunately, intrinsic and acquired resistance limits the effectiveness of these drugs (17). Among the mechanisms that account for the resistance of lung tumors to EGFR inhibitors is the presence of somatic mutations in genes encoding components of the growth factor signaling such as RAS, a GTPase that acts as a signaling molecule downstream of EGFR (19). RAS and EGFR mutations are mutually exclusive in NSCLC (19). Of note, 90% of RAS mutations in lung adenocarcinoma represent alterations in K-RAS and most of them involve single substitutions at residues 12 and 13 (20). Certain K-RAS mutations have been associated with worse prognosis in lung cancer (oncogenic mutations in G12 residues) and worse response of NSCLC and colorectal cancer to treatment with EGFR–TKIs and the EGFR-directed antibodies, respectively (19, 21, 22).

Multiple clinical and laboratory-based studies have demonstrated antiproliferative and proapoptotic functions for ERβ in various types of cancers, including breast, colon, prostate, and ovarian cancer (6). In contrast, the role of ERβ in lung cancer development and progression is still poorly understood. In this study, we investigated whether ERβ elicits antitumorigenic actions in NSCLC cells that may account for the correlation between ERβ and better survival observed in patients with NSCLC.

Adenocarcinoma A549 and large-cell carcinoma H1299 and H661 NSCLC cell lines were obtained from the American Type Culture Collection. H1299 and H661 cells were cultured in RPMI-1640 and A549 cells in Dulbecco's Modified Eagle Medium media supplemented with 10% FBS. In the experiments with ligand treatment, cells were cultured in phenol red–free media containing 0%, 2% or 5% dextran-coated charcoal (DCC)–treated FBS. Cells were treated with either 10 nmol/L 17β-estradiol (E2) or 10 nmol/L 5α-androstane-3β,17β-diol (3β-adiol). ERβ expression constructs were generated by cloning the full-length ERβ1 or the C-terminally truncated isoform ERβ2 (also known as ERβcx) in the pIRESneo3 expression vector (Clontech). H1299 and A549 cells were stably transfected with an empty pIRESneo vector or the recombinant pIRESneo-ERβ1 or pIRESneo-ERβ2 plasmids. H661 cells were infected with lentivirus containing the empty plenti6/V5 vector or the recombinant pLenti6/V5-D-FLAG-ERβ1 plasmid as previously described (23). H1299 cells were transiently transfected twice with ERβ-specific siRNAs (Invitrogen) target sequences 1# 5′-TAGCGACGTCTGTCGCGTCTTCAC-3′, 2# 5′-TATTGACCGCTACCTGGTGATTTCC-3′. An siRNA-targeting luciferase was used as a control (Cat. no. 12935-146; Invitrogen). For the expression of wild-type EGFR, H1299 cells were stably transfected with the pBABE–EGFR plasmid (Addgene; plasmid #11011) using the empty pBABE vector as a control (Addgene; plasmid # 1764). A plasmid with mutant N-RAS(61K) was also purchased from Addgene (plasmid # 12543). The coding sequence of mutant N-RAS was subcloned into the pIRESpuro3 vector (Clontech). The myc epitope (underlined portion in the reverse primer) was tagged on the C-terminus of mutant N-RAS using the following primers: FW 5′- GTACGACCGGTGCCACCATGACTGAGTACAAACTG GT-3′ and RV 5′-AGCAGGATCCTTACAGATCTTCTTCAGAAATAAGTTTTTGTTCCATCAACACATGGCA-3′. ERβ1-expressing H1299 cells were stably transfected with an empty pIRESpuro vector or the pIRESpuro-N-RAS(61K) plasmid.

Following incubation in 5% DCC–FBS media for 48 hours, cells were harvested by trypsinization and replated at the density of 1 × 10³ cells per 60-mm dish in triplicates and were either mock (EtOH) treated or treated with E2. After 14 days, cells were washed, fixed in an 3:1 methanol:acetic acid solution, and stained with 0.5% crystal violet in 25% methanol. Surviving colonies were counted in each well and the plating efficiency was calculated using the equation plating efficiency = number of colonies counted/number of cells plated × 100. Then, the fraction of the cells surviving the expression of ERβ1 and/or the treatment with E2 was determined by normalizing the plating efficiency of the ERβ1-expressing and/or E2-treated cells to that of the plates with control untreated cells, which was set to 100%.

H1299 and H661 cells were seeded onto 12-well plates at a density of 1 × 10⁴ cells per well in 5% DCC–FBS media. Cells were incubated for 5 days with EtOH or E2. Cells were then fixed in ice-cold methanol and stained with 0.1% crystal violet in PBS. Plates were air-dried and the stained cells were solubilized in 10% SDS solution overnight. The optical density of the extracted dye was measured with a spectrophotometer at 590 nm. Optical density measurements were used to generate proliferation curves.

After treatment, H1299 and A549 cells were harvested by trypsinization and fixed in 70% ethanol overnight at −20°C. Cells were resuspended in 50 μg/mL propidium iodide and analyzed on a BD FACSAria II cell sorter. The data were analyzed using a 7.0 FlowJo software.

A549 cells were plated onto 96-well plate at a density of 1 × 10⁴ per well in triplicates. After treatment with EtOH or E2 for 24 hours, cells were washed with PBS, lysed and the luminescent signal generated by the cleavage of the proluminescent caspase-3/7 substrate, that is, proportional to the caspase-3/7 activity, was measured according to the manufacturer's protocol (Promega).

Total RNA was isolated using the Aurum Total RNA Mini Kit (Bio-Rad) and reverse transcribed to cDNA using the iScript cDNA Synthesis Kit (Bio-Rad). Real-time quantitative reverse transcription PCR (qRT-PCR) was performed using the iTaq SYBR Green Kit (Bio-Rad). All quantitative data were normalized to GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and 36B4. The primer sequences for CDKN1B are 5′-CCCTTTCAGAGACAGCTGATAC-3′ and 5′-CACCAGATCTCCCAAATGAGAA-3′, for GAPDH 5′-TGATGACATCAAGAAGGTGGTGAAG-3′ and 5′-CCTTGGAGGCCATGTGGGCCAT-3′, for 36B4 5′-GCAATGTTGCCAGTGTCTGT-3′ and 5′-GCCTTGACCTTTTCAGCAAG-3′.

Cells were lysed in radioimmunoprecipitation assay buffer (50 mmol/L Tris-HCl, pH 7.5, 150 mmol/L NaCl, 0.1% SDS, 0.5% deoxycholate, and 1% NP-40) containing protease (1 mmol/L EDTA, Roche protease inhibitor mixture, and 2 mmol/L phenylmethylsulfonylfluoride) and phosphatase inhibitors (1 mmol/L NaF, 1 mmol/L Na3VO4, and Sigma phosphatase inhibitor mixture). Nuclear and cytoplasmic extracts were prepared as described before (24). The lysates were subjected to SDS–PAGE and the proteins were transferred onto nitrocellulose membranes. Membranes were blocked with 5% nonfat milk in TBST (0.05% Tween-20) for 3 hours at room temperature and probed with the primary antibodies overnight at 4°C. Primary antibodies against ERβ1 (clone 14C8) and p84 were purchased from Genetex. For validation purposes two additional ERβ antibodies were used. These include a monoclonal anti-ERβ antibody (clone 68-4; Millipore) and an anti-ERβ C-terminus rabbit polyclonal antibody that recognizes only the ERβ1 isoform (Invitrogen; Supplementary Fig. S1). The primary antibodies for EGFR, phospho-ERK (extracellular signal–regulated kinase)1/2, caspase-3, p27, p21, α-tubulin, and c-MYC were purchased from Santa Cruz Biotechnology and the antibodies against BIM, ERK1/2, pAKT(S473), pan-AKT, RAS, cyclin D2, and cleaved caspase-3 were from Cell Signaling Technology. β-Actin and Flag antibodies were obtained from Sigma. The membranes were incubated with secondary antibodies for 2 hours at room temperature. Proteins were visualized using ECL detection reagents (Amersham Biosciences). Band intensities were quantified by densitometry using the ImageJ software.

RAS activity was assessed using a Pan-RAS Activation Kit from Cell Biolabs. The active RAS–GTP form interacts with the effector RAF-1. The RAS–GTP was pulled down from cell lysates using the RAS-binding domain of RAF-1 immobilized to agarose beads. The levels of precipitated RAS–GTP were measured by immunoblotting using an anti–Pan-RAS antibody (Cell Biolabs).

Oncogenic mutations that frequently occur in lung cancer are attractive targets for anticancer therapy. Although a few targets like EGFR have been successfully targeted, direct inhibition of some mutant genes, such as K-RAS remains elusive. Patients with K-RAS mutations tend to have poor prognosis and do not respond to EGFR–TKIs (25, 26). Interestingly, a positive correlation between the expression of ERβ1 and better survival has been observed in patients with NSCL tumors that carry EGFR mutations (9, 10). However, the molecular basis for this correlation is unknown. We hypothesized that ERβ1 regulates cell survival in NSCLC and that high expression of ERβ1 in NSCLC cells is associated with decreased cell proliferation and induction of apoptosis. To test this hypothesis, we stably expressed ERβ1 in three NSCLC cell lines that express very low levels of endogenous ERβ1 (Fig. 1C). These include the H1299 and A549 cells that carry N-RAS and K-RAS mutations, respectively, and the H661 cells that express wild-type RAS. To achieve comparable expression of ERβ1 in the cell lines, A549 and H1299 cells were stably transfected with the pIRES-ERβ1 plasmid, whereas the H661 cells that proved difficult to transfect were stably infected with lentivirus containing the pLenti-FLAG-ERβ1 plasmid (Fig. 1C). As shown in Supplementary Fig. S1D, ERβ1 was localized in the nucleus of the NSCLC cells. Induction of ERβ1 expression inhibited cell growth in all three NSCLC cell lines (Fig. 1A and Supplementary Fig. S2). Cell survival assays revealed that ERβ1 decreased cell growth even in the absence of ligand and that addition of E2 in cell culture further enhanced the cytotoxic effect of ERβ1 in H1299 but not in A549 and H661 cells (Fig. 1A and Supplementary Fig. S2). The ligand-independent antitumorigenic function of ERβ1 that was observed in NSCLC cells has also been previously described in breast, colon, and prostate cancer cells (23, 24, 27, 28). Interestingly, the effect of ERβ1 was more potent in H1299 and A549 cells that express mutant RAS compared with H661 cells that carry wild-type RAS, suggesting that ERβ1 may suppress the growth of NSCLC cells by targeting oncogenic RAS signaling (Fig. 1A). To elucidate the mechanism through which ERβ1 inhibited NSCLC cell growth, control and ERβ1-expressing cells were analyzed for cell-cycle progression and apoptosis by flow cytometry. As shown in flow cytometry histograms in Fig. 1B, upregulation of ERβ1 inhibited cell-cycle progression by arresting the H1299 cells in G₁ phase. In addition to G₁–S phase cell-cycle arrest, ERβ1 enhanced apoptosis as shown by the higher percentage of ERβ1-expressing H1299 and A549 cells in sub-G₁ fraction that is indicative of apoptosis (Fig. 1B). These results suggest that upregulation of ERβ1 in NSCLC cells inhibits cell growth by inducing cell-cycle arrest and apoptosis.

To investigate the molecular mechanism involved in the ERβ1-induced cell-cycle arrest and apoptosis, control and ERβ1-expressing H1299 and A549 cells following treatment with E2 or 3β-adiol for 24 hours were initially assessed for the expression of the cell-cycle regulatory proteins p21(WAF/CIP1) and p27(KIP1) that retard cell-cycle progression by inhibiting the activity of cyclin-dependent kinases (29). We treated the cells with the same concentration of the androgen metabolite 3β-adiol. 3β-Adiol that displays high affinity for ERβ has been shown to elicit antiproliferative effects in prostate cancer and to affect cell proliferation in other reproductive tissues (30). Although expression of the enzymes 5α-reductase and 3β-hydroxysteroid dehydrogenase that participate in the generation of 3β-adiol has been described in NSCLC cells and tissues, its presence and role in lung cancer is still poorly understood (31). As shown in Fig. 2A and B, ERβ1-expressing H1299 and A549 cells had higher levels of p21 and/or p27 compared with control cells. In line with the immunoblotting results, gene expression analysis by quantitative RT-PCR (qRT-PCR) indicated that ERβ1 upregulates p27 mRNA levels (Fig. 2C). This is consistent with previous studies showing upregulation of p21 and/or p27 by ERβ1 in breast, prostate, and colon cancer cells (23, 32, 33). In addition to cell-cycle inhibitors, proapoptotic factors were analyzed in control and ERβ1-expressing cells. Caspase-3 that is activated by proteolytic cleavage is one of the executioner caspases of the intrinsic apoptotic pathway (34). To examine whether the intrinsic apoptotic pathway is involved in the ERβ1-induced apoptosis, the cleavage of caspase-3 was evaluated in control and ERβ1-expressing H1299 and A549 cells. Immunoblotting and a luciferase-based assay for the assessment of caspase-3/7 activity revealed increased cleavage and activity of caspase-3 in two ERβ1-expressing NSCLC cell lines, both in the absence and presence of the ligands E2 or 3β-adiol (Fig. 2A and B, top and bottom). As shown in Fig. 2A and B, although no significant differences in the levels of cleaved caspase-3, p21, and p27 were observed between E2- and 3β-adiol–treated cells, E2 treatment caused downregulation of ERβ1 to a higher extent than 3β-adiol. This ligand-mediated downregulation of ERs has been associated with the activation of the receptors (35, 36). Importantly, in contrast with ERβ1, stable expression of the ERβ splice variant ERβ2, that differs from ERβ1 in 26 C-terminal amino acids, failed to significantly decrease cell growth and it did not enhance apoptosis in H1299 cells (Fig. 2D and Supplementary Fig. S1C). This suggests that the proapoptotic phenotype observed in ERβ1-expressing cells was due to specific upregulation of the fully functional ERβ1. These results indicate that ERβ1 inhibits the growth of NSCLC cells by upregulating cell-cycle inhibitors and stimulating the intrinsic apoptotic pathway.

EGFR signaling is associated with the progression and resistance of NSCLC to targeted therapy (17). EGFR and downstream components activate prosurvival signaling pathways mainly by regulating protein abundance of members of the BCL-2 family of proapoptotic factors (37). Although H1299 and A549 NSCLC cells do not carry EGFR mutations, they overexpress an active form of the receptor as a result of constitutive secretion of EGFR ligands (16). We hypothesized that ERβ1 induces apoptosis in NSCLC cells by regulating EGFR signaling. Immunoblotting analysis revealed reduced expression of EGFR and decreased activity of ERK1/2 that act downstream of EGFR in ERβ1-expressing cells compared with the control cells (Fig. 3A and B). This effect was observed in the absence of ligands and the levels of EGFR and phosphorylated ERK1/2 were similar between E2- and 3β-adiol–treated cells. ERK1/2 are components of the prosurvival signaling pathway that inhibits apoptosis by promoting proteasomal degradation of the proapoptotic BCL-2 family member BIM (38). As expected, increased expression of BIM was found in ERβ1-expressing cells compared with the control cells (Fig. 3A). Furthermore, inhibition of ERK1/2 has been reported to impair the production of EGFR ligands (39). As shown in Fig. 3C, assessment of the expression of the intrinsic ligand of EGFR, EGF, by qRT-PCR revealed reduced EGF mRNA levels in ERβ1-expressing cells. To strengthen our results connecting the proapoptotic phenotype in NSCLC with the specific upregulation of ERβ1, ERβ1 was depleted in ERβ1-expressing H1299 cells by siRNA knockdown. Immunoblotting showed that downregulation of ERβ1 using two specific ERβ siRNAs rescued ERK1/2 phosphorylation and decreased BIM expression and caspase-3 cleavage (Fig. 3D). Taken together, these results suggest that ERβ1 stimulates proapoptotic pathways in NSCLC cells by inactivating EGFR signaling.

To investigate whether repression of EGFR signaling is essential for the inactivation of ERK1/2 and the enhanced apoptosis observed in ERβ1-expressing NSCLC cells, control and ERβ1-expressing H1299 cells were treated with EGF and analyzed for the activity of factors downstream of EGFR signaling. Upon ligand binding, EGFR is activated and through its interaction with the adaptor proteins GRB2-associated–binding protein 1 (GAB1) and growth factor receptor–bound protein 2 (GRB2) activates the PI3K (phosphoinositide 3-kinase)–AKT and the ERK pathways, respectively. Recruitment of GRB2 to EGFR results in activation of the RAS–RAF signaling cascade, which in turn activates ERK1/2 (40). As expected, treatment of H1299 cells with EGF rapidly induced EGFR phosphorylation at Tyr1068, which is indicative of EGFR activation and significantly increased the activity of AKT as shown by its increased phosphorylation at S473 in both control and ERβ1-expressing cells (Fig. 4A). In contrast, EGFR activation did not reverse the ERβ1-mediated decrease in ERK1/2 phosphorylation, suggesting that ERβ1 decreases the activity of ERK1/2 by acting on a component downstream of EGFR. To confirm this, we overexpressed EGFR in ERβ1-expressing H1299 cells. As shown in Fig. 4B, EGFR overexpression significantly increased the phosphorylation of AKT but did not reverse the ERβ1-mediated inactivation of ERK1/2 and the increased expression of BIM. Similarly, treatment of ERβ1-expressing cells that overexpress EGFR with EGF although profoundly increasing the levels of phosphorylated AKT at S473, failed to significantly increase the activity of ERK1/2 (Fig. 4C). Consistent with the effect on ERK signaling, overexpression of EGFR failed to rescue the H1299 cells from the ERβ1-induced apoptosis strengthening our hypothesis that ERβ1 induces apoptosis by inhibiting growth factor signaling downstream of receptor tyrosine kinases (Fig. 4D).

Downstream of EGFR, RAS activates ERK1/2 by interacting with and regulating the activity of RAF. RAS is one of the most frequently mutated genes in NSCLC. Of note, 90% of RAS mutations in lung adenocarcinoma represent alterations in K-RAS and K-RAS mutations correlate with a worse prognosis in lung cancer and are implicated in the resistance to EGFR–TKIs (19). We investigated whether RAS is involved in the ERβ1-mediated regulation of EGFR and ERK1/2. We overexpressed mutant N-RAS in ERβ1-expressing H1299 cells and, following incubation in complete media or media lacking growth factors, we analyzed these cells as well as control and ERβ1-expressing cells for EGFR expression and ERK1/2 phosphorylation. Upregulation of mutant N-RAS was found to rescue EGFR for the ERβ1-mediated downregulation, suggesting that the decreased expression of EGFR in ERβ1-expressing cells was RAS dependent. This effect was more potent in the presence of growth factors, suggesting a ligand-mediated regulation of EGFR by RAS and ERβ1 (Fig. 4A and Fig. 5A, top). Previous studies have reported increased endocytic trafficking and degradation of EGFR by the downstream AKT in the presence of EGF (40). In addition, we have previously shown increased degradation of EGFR in ERβ1-expressing triple-negative breast cancer cells in the presence of EGF (24). Taken together, these results suggest that ERβ1 may induce degradation of EGFR in NSCLC cells by regulating the activity of the downstream RAS. In contrast with restoring EGFR levels, overexpression of mutant N-RAS did not reverse the ERβ1-mediated inactivation of ERK1/2, suggesting that ERβ1 may block the activity of RAS (Fig. 5A, top). Indeed, as shown in the bottom of Fig. 5A, ERβ1 decreased the activity of RAS in H1299 cells, suggesting that ERβ1 may elicit tumor-suppressive actions in NSCLC cells by inactivating mutant RAS.

Several factors have been shown to facilitate the maintenance of RAS-dependent tumors (25). Inhibition of the transcription factor MYC triggers rapid regression of mutant RAS-induced tumors in vivo (41). We investigated whether ERβ1 affects c-MYC, which in response to diverse extracellular and intracellular signals acts downstream of RAS/ERK to promote cell growth. As shown in Fig. 5B and C, upregulation of ERβ1 decreased the expression of c-MYC and that of the c-MYC target gene cyclin D2 in H1299 and A549 cells both in absence and presence of ERβ ligands. Similarly to the expression of cleaved caspase-3, no difference was observed in the levels of cyclin D2 between E2- and 3β-adiol–treated cells. As in the case of RAS and ERK1/2 activity, EGFR upregulation in ERβ1-expressing H1299 cells did not affect the expression of c-MYC strengthening our findings that EGFR downregulation is not the critical event in the ERβ1-mediated inhibition of the signaling that stimulates cell growth (Fig. 5D). These results demonstrate that ERβ1 downregulates effectors of the RAS/ERK pathway in NSCLC cells by reducing the activity of RAS independently of EGFR.

The discovery of oncogenic mutations in NSCLC has improved the knowledge of the aberrant molecular signaling found in this lung cancer subtype and led to the development of biomarkers with associated targeted therapeutics (18). Although EGFR mutations and anaplastic lymphoma kinase translocations were successfully targeted with EGFR–TKIs and crizotinib, respectively, direct blockade of mutant K-RAS, which accounts for about 30% of all mutations in lung adenocarcinoma, remains inefficient (26, 42). In addition to the overactive growth factor signaling, dysregulation of other pathways that regulate cell growth such as those mediated by ERs has been linked to lung cancer development and progression. Interestingly, lower levels of circulating estrogen in women with lung cancer over the age of 60 correlated with better survival and hormone replacement therapy has been associated with shorter median survival (2, 43). The adverse effects of estrogen could be mediated by either ERα or ERβ because both ERs are expressed in lung tumors according to the profile of NCBI EST databases and studies that analyzed ER protein expression in human lung cancers (8, 44). However, the correlation of ERβ1 with better outcome and that of ERα with worse survival and poorer prognosis in patients with NSCLC suggest that ERα, that promotes cell proliferation in breast cancer, may also mediate the tumorigenic actions of estrogen in lung tissue (6, 8–10, 45, 46). In addition to the protective function proposed in lung cancer, ERβ1 is known to inhibit the growth of breast, ovarian, colon, and prostate cancer cells (6). Although treatment of NSCLC cells with ERβ agonists has been reported to stimulate cell proliferation, the role of ERβ in regulating cell survival and apoptosis in lung cancer still remains unclear (7, 12).

In this study, we carried out experiments to determine functions of ERβ1 in NSCLC cells that may account for its association with the better clinical outcome of patients with NSCLC. Given that mutant RAS correlates with worse prognosis and is implicated in the acquired resistance to EGFR–TKIs, we investigated the role of ERβ1 in regulating cell survival in NSCLC that express wild-type and mutant RAS (25). Immunoblotting analysis, based on the use of appropriate controls and different ERβ antibodies that had been previously validated for their specificity, revealed that the NSCLC cells we studied express very low (no detectable) levels of ERβ1 (Fig. 1C and Supplementary Fig. S1; refs. 24, 28). Induction of ERβ1 expression in these cells profoundly decreased cell growth. The growth inhibitory effects of ERβ1 were mostly observed in the absence of ligand, which is consistent with previous studies demonstrating ligand-independent antitumorigenic actions of ERβ1 in different types of cancer cells (23, 24, 27, 28, 47). However, treatment with E2 further suppressed cell growth in one of the NSCLC cell lines. The same treatment caused downregulation of ERβ1 that has been associated with receptor activation (35, 36). This ligand-dependent ERβ1-mediated regulation of cell survival that was observed only in survival assays after long-term treatment of the cells with E2 may suggest the use of specific ERβ1 agonists as potential treatment modality for the clinical management of NSCLC. Further analysis of the NSCLC cells revealed that ERβ1 induces G₁–S cell-cycle arrest and apoptosis by increasing the levels of the proapoptotic marker cleaved caspase-3 and the cell-cycle inhibitors p21 and p27.

Interestingly, the cell growth inhibitory effects of ERβ1 were more potent in NSCLC cells that express mutant RAS, suggesting the involvement of RAS and growth factor signaling in the ERβ1-mediated regulation of NSCLC cell survival. Upon growth factor binding, EGFR is activated and, through binding the GRB2, promotes the recruitment of guanine nucleotide exchange factors (GEF) to the plasma membrane in which RAS is localized as a result of farnesylation. The increased interaction of GEFs with RAS facilitates the formation of the active GTP-bound state RAS and the subsequent activation of the downstream ERK pathway (17). We investigated whether inactivation of the EGFR/RAS/ERK signaling axis was associated with the induction of apoptosis in ERβ1-expressing cells. Indeed, decreased protein levels of EGFR and activity of ERK1/2 were detected in ERβ1-expressing NSCLC cells. These results together with the upregulation of the proapoptotic marker BIM that predicts response to EGFR–TKI treatment and is degraded in response to ERK1/2 activation strengthened the inhibition of the EGFR–RAS pathway by ERβ1 in NSCLC cells (38, 39, 48).

Importantly, knockdown of the transfected ERβ1 in NSCLC cells reversed the proapoptotic phenotype as shown by the decreased levels of the proapoptotic cleaved caspase-3 and BIM and restored ERK1/2 activity. In addition, upregulation of the ERβ splice variant ERβ2 did not significantly affect cell growth and apoptosis in NSCLC cells, strengthening the association of the proapoptotic phenotype and the inhibition of the growth factor signaling in NSCLC cells with the specific upregulation of ERβ1. The expression of ERβ2 has been associated with various clinical outcomes in cancer. In particular, it has been correlated with increased survival and invasiveness of prostate and ovarian cancer cells. In breast cancer, nuclear ERβ2 has been associated negatively with metastasis and cytoplasmic ERβ2 with worse outcome (49, 50). ERβ2 has been suggested to elicit its biologic functions by modulating the transcriptional activity of wild-type ERα and ERβ through heterodimerization or by interacting with the membrane and cytoplasmic signaling cascade (6). Upregulation of ERβ2 had no significant impact on the survival of NSCLC cells that do not express wild-type ERα and ERβ. However, it might differentially affect the phenotype of NSCLC cells that coexpress ERα and ERβ1 by modulating their activity. In such cellular context, in contrast with ERβ1, ERβ2 may increase the survival and metastatic potential of NSCLC cells and tumors.

To provide more insights into the mechanism through which ERβ1 regulates the EGFR–RAS pathway in NSCLC cells, we modified the expression of EGFR and mutant RAS in ERβ1-expressing cells. Restoring EGFR activity by EGF treatment or EGFR upregulation in ERβ1-expressing NSCLC cells, although increasing the activity of the PI3K/AKT pathway, failed to reverse the ERβ1-mediated downregulation of phospho-ERK1/2 and the subsequent upregulation of BIM and enhanced apoptosis. This suggests that direct blockade of oncogenic RAS downstream of EGFR and not downregulation of EGFR may be essential for the apoptosis observed in ERβ1-expressing NSCLC cells. Indeed, induction of ERβ1 expression decreased the activity of RAS and upregulation of mutant RAS in ERβ1-expressing cells reversed the EGFR downregulation indicating the central role of oncogenic RAS inhibition in ERβ1-mediated phenotype in NSCLC cells, including the regulation of EGFR (Fig. 6). The downregulation of EGFR by ERβ1 in the presence of EGF suggests that ERβ1 may induce degradation of EGFR by regulating the activity of downstream RAS. This is consistent with previous studies showing increased degradation of EGFR in cells with altered activity of the downstream AKT (40). The inhibition of oncogenic RAS by ERβ1 in lung cancer cells was further supported by the downregulation of the effector of the RAS/ERK signaling c-MYC and its direct target cyclin D2 independent of EGFR in ERβ1-expressing NSCLC cells (41). In addition to lung cancer, certain K-RAS mutations have been associated with worse prognosis in colorectal cancer (19, 21, 22). Interestingly, ERβ1 has been shown to inhibit the proliferation of the mutant RAS SW480 colon adenocarcinoma cells by downregulating c-MYC and increasing the expression of p27 (23). Given that the expression of both factors was altered in ERβ1-expressing NSCLC cells in which the activity of RAS decreased, it is possible that an oncogenic RAS inactivation could account for the ERβ1-mediated tumor repressive functions in colon cancer cells.

Although ERβ1 has been previously reported to inhibit the growth of breast, colon, ovarian, and prostate cancer cells, this is the first demonstration that ERβ1 decreases lung cancer cell survival by regulating oncogenic RAS (23, 24, 27, 28). These results may shed more light into the mechanisms that regulate resistance to targeted therapy in lung cancer cells and explain the association between ERβ1 and outcome patients with of NSCLC observed in clinical studies. Further understanding of the mechanisms that suppress oncogenic RAS and decrease cell survival in ERβ1-expressing cells is necessary to establish ERβ1 as a tumor suppressor in NSCLC and as a factor with potential utility in the prognosis and treatment of the disease.

J.-A. Gustafsson is a senior professor at Karolinska Institutet and has ownership interest (including patents) in Karo Bio AB. No potential conflicts of interest were disclosed by the other authors.

Conception and design: F. Nikolos, C. Thomas, J.-Å. Gustafsson

Development of methodology: F. Nikolos, C. Thomas, G. Rajapaksa, I. Bado

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Nikolos, C. Thomas

Writing, review, and/or revision of the manuscript: F. Nikolos, C. Thomas, J.-Å. Gustafsson

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Thomas, J.-Å. Gustafsson

Study supervision: C. Thomas, J.-Å. Gustafsson

This study was supported by grants from the Emerging Technology Fund of Texas, CPRIT, the Welch Foundation (E-0004), and the Swedish Cancer Society.

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.