This study was approved by the Institutional Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology (IRB ID no. 20141101). Paired samples of primary NSCLC tumors and corresponding non-tumorous lung tissues from 241 Chinese patients were obtained at the time of surgical resection at the Department of Thoracic Surgery, Affiliated Tongji Hospital of Huazhong University of Science and Technology Tongji Medical College (Wuhan, China) from August 2013 to September 2015. None of the patients underwent chemotherapy or radiotherapy before surgery. Patient demographics, including gender, age, smoking history, pathological diagnosis, pathological tumor-node-metastasis stage (18), and tumor differentiation grade, were obtained from the Tongji Hospital records. Baseline characteristics of the patients are documented in Table I.

Among the 241 patients, metastatic lymph nodes were obtained from 30 patients with paired primary tumors; metastatic lymph nodes were obtained via surgical resection of the primary tumor with lymph node dissection. Patients with lymphadenitis and primary malignancies of the lymph node were excluded. Details on the 30 cases with IIA-IIIB NSCLC with lymph node metastasis have been provided in a previous study from our group (6). Palliative care or surgical biopsy was administered, after obtaining informed consent, to 32 patients with inoperable stage IIIb–IV primary NSCLC.

Fresh tissues were immediately snap-frozen and stored at −80°C, or formalin-fixed and embedded in paraffin. Samples were diagnosed and confirmed by at least two lung cancer pathologists. TMA was prepared by Outdo Biotech Co., Ltd. (Shanghai, China). All patients had provided their written, informed consent to tissue collection prior to sampling, and all procedures involving human samples were conducted in accord with the Declaration of Helsinki.

Sample processing and immunohistochemistry were performed as previously described (6). Rabbit anti-human TLR4 polyclonal antibody (Ab) (dilution 1:100, cat. no. ab13556) and rabbit anti-human ERβ monoclonal Ab (dilution 1:100, cat. no. ab3577) were purchased from Abcam. Protein expression levels were scored independently by two pathologists. Immunoreactivity scores of cancer-tissue samples were determined based on staining intensity and area of positive staining according to the method described in Tang et al (19). Positive cells were scored as follows: 1, ≤20% positive cells; 2, 20–50% positive cells; 3, 50–75% positive cells; and 4, >75% positive cells. Staining intensity was evaluated as follows: 1, negative; 2, weakly positive; 3, moderately positive; and 4, strongly positive. A score of 1–16 was obtained by multiplying the staining intensity and proportion of positive cells: (−), ≤4; (+), >4 and ≤; (++), >8 and ≤ 12; and (+++), >12 and ≤16. A total score >12 was defined as high expression, and a score ≤8 was defined as low expression.

Human NSCLC cell lines PC9, A549, H1793 and H1975 were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA), grown for 2 weeks, and passaged four times before freezing aliquots for subsequent analyses. The cell lines were tested and authenticated by ATCC. The normal human bronchial epithelial (HBE) cell line was obtained from the Shanghai Cancer Institute (Shanghai, China). PC9, A549, H1975 and HBE cells were cultured in Roosevelt Park Memorial Institute Medium (RPMI)-1640 medium, and H1793 cells were cultured in Dulbecco's modified Eagle's medium (DMEM):nutrient mixture F12. All media (HyClone; GE Healthcare) were supplemented with 10% fetal bovine serum (FBS; Clark Bioscience). Cells were incubated in a humidified atmosphere with 5% CO₂ and 95% air at 37°C.

The expression vectors and small interfering RNAs (siRNAs) targeting ERβ were constructed as previously described (6,20). Briefly, TLR4- or ERβ-expressing plasmids (pcDNA3.1-TLR4 or -ERβ) and corresponding empty plasmids were obtained from OriGene Technologies, Inc. siRNAs targeting TLR4 or ERβ (TLR4-siRNA or ERβ-siRNA Stealth siRNAs HSS103378; Life Technologies) and control siRNAs (ctrl-siRNA or siRNA-NC; Life Technologies) were constructed using a gene silencing vector. Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Inc.) was used during transfection following the manufacturer's instructions. TLR4 and ERβ protein expression was analyzed by western blotting following transfection with plasmids or siRNAs.

Parental NSCLC cells and mice in the metastatic model were exposed to E2 (Sigma-Aldrich; Merck KGaA), 4,40,400-(4-propyl-[1H]-pyrazole-l,3,5-triyl), LPS (isolated from Escherichia coli 0111:B4; Sigma-Aldrich; Merck KGaA), fulvestrant (Ful; an ER antagonist; Cayman Chemical), and TAK-242 (also known as CLI-095; InvivoGen), either alone or in combination. The dosages of each drug, used in vitro and in vivo, were used as previously (21–24). Each group of cells was treated for 48 h and harvested for further analysis. Cell culture experiments were performed using reagents dissolved in 100% dimethyl sulfoxide (DMSO), and DMSO was also used as a control (vehicle).

Western blotting was performed as previously described (6). The primary antibody (Ab) used for western blots included rabbit anti-human ERβ (dilution 1:1,000) from Abcam (cat. no. ab3577), rabbit anti-human TLR4 (dilution 1:1,000) from Abcam (cat. no. ab13556), mouse anti-human MMP-2 (dilution 1:500) from Santa Cruz Biotechnology (cat. no. sc-53630), rabbit anti-human P38MAPK (dilution 1:500, cat. no. AP0424)/p-P38MAPK (dilution 1:500, cat. no. BS4844), and rabbit anti-human tAkt (dilution 1:500, cat. no. AP0485)/p-Akt (dilution 1:500, cat. no. AP0484) from Bioworld Technology, rabbit anti-human P65NF-κB (dilution 1:1,000, cat. no. ab16502)/p-P65NF-κB (dilution 1:1,000, cat. no. ab86299) from Abcam, rabbit anti-human myd88 (dilution 1:1,000) from Proteintech (cat. no. 23230-1-AP, and mouse anti-human GAPDH (dilution 1:10,000) from Cell Signaling Technology (cat. no. 51332). The densitometry of the western blots was analyzed by ImageLab software (version 6.0.0; Bio-Rad Laboratories, Inc.).

Cultured A549 cells were placed into 10-cm dishes and transfected with empty vector pcDNA3.1-ERβ, pcDNA3.1-TLR4, siRNA-ERβ, or siRNA-TLR4. The cells were then washed in 1X phosphate-buffered saline (PBS) and resuspended in 1 ml of NP-40 lysis buffer (70 mM NaCl, 50 mM Tris pH 8, and 0.5% NP-40) supplemented with phosphatase inhibitor and protease inhibitor cocktails (Sigma-Aldrich; Merck KGaA), and the cell lysates were ultrasonicated. After rotating at 4°C for 30 min, the cell lysates were collected and precleared by centrifugation at 12,000 rpm (12,800 g rcf) at 4°C for 10 min. Protein concentration was assessed using 1 mg of total protein via the Bradford Assay (Bio-Rad Laboratories, Inc.). For co-immunoprecipitation of endogenous ERβ with TLR4, total protein was isolated from A549 cells as described previously (22). For each pull-down, 5 µg of anti-ERβ Ab (dilution 1:200, cat. no. ab3577; Abcam) or anti-TLR4 Ab (dilution 1:200, cat. no. ab30667; Abcam) was added to the normalized lysate, and the mixture was incubated overnight at 4°C. Normal rabbit IgG (Santa Cruz Biotechnology) was used as the negative control. Immune complexes were then precipitated with protein A/G plus agarose (Thermo Fisher Scientific, Inc.). Immunoprecipitates were collected by centrifugation (2,850 × g rcf, 4°C, 3 min) and washed gently with lysis buffer. Immunoprecipitated samples were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS/PAGE) for immunoblotting.

NSCLC cell lines A549 and H1973 were cultured in a 6-well culture plate until 90% confluence. The cell monolayers were then scratched using a 200 µl sterile pipette tip. Cell migration was determined by measuring the movement of cells into the scratched area. Representative images (×40) of wound closure were captured at 0 and 24 h using an inverted light microscope (Olympus Corp.).

Transwell^® Permeable Supports (inserts 6.5 mm in diameter; Corning, Inc.) were used for the cell migration assay. For the cell invasion assay, Transwell^® Permeable Support inserts were coated with BD Matrigel™ Basement Membrane Matrix (BD Biosciences). Cells (A549 and H1793) suspended in serum-free media were added to the upper chamber at various densities depending on the cell line. Migration and invasion were evaluated based on the number of cells invading the Transwell^® membrane, and counting was performed using an Olympus microscope (Olympus Corp.) at ×100 magnification. Four fields were randomly selected for analysis. Detailed procedures are described elsewhere.

After 48 h of treatment with specific drugs or combinations of drugs, A549 and H1793 cells were seeded overnight on coverslips in a 24-well plate. After the cells reached 50% confluence, they were washed in 1X PBS, fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.1% Triton X-100 for 15 min, blocked in 5% goat serum at room temperature (RT) for 30 min, and incubated overnight at 4°C with the indicated primary Ab. Next, cell-seeded coverslips were rewarmed for 1 h and incubated with a specific secondary Ab for 2 h at 37°C. Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) for 5 min at RT. Next, the samples were washed in 1X PBS for 5 min four times. Finally, the coverslips were observed under a fluorescence microscope (Olympus Corp.; magnification of ×200).

The 3D invasion assay was described previously (25). The Cultrex^® 3D Spheroid Cell Invasion Assay kit (Trevigen) was utilized in this procedure. Briefly, 1×10⁵ cells in 500 ml fresh culture media containing 2.5% Matrigel and 5 ng/ml spheroid formation extracellular matrix (ECM) were plated into 24-well plates coated with a collagen/Matrigel mixture. The spheres with protrusions were considered positive for cell invasion. Distances between the invasive cell frontier and the spheroid edge were measured on day 1, 3 and 6 using an Olympus IX70 (Olympus) inverted microscope (magnification of ×100). Each experiment was repeated twice, and each procedure was performed in triplicate.

Coverslips were cleaned with 20% nitric acid and coated with poly-L-lysine in a 24-well plate. Poly-L-lysine was fixed with 0.5% glutaraldehyde before adding fluorescein isosthiocynanate-conjugated (FITC) gelatin (QCM™ Gelatin Invadopodia Assay kit (Green), EMD Millipore). A thin layer of FITC-conjugated gelatin was placed onto the coverslips and was crosslinked using glutaraldehyde on ice for 10 min. Crosslinking was continued at RT for an additional 30 min. The coverslips were rinsed with PBS, incubated with 5 mg/ml sodium borohydride at RT for 3 min, rinsed again with PBS, incubated with 70% ethanol for 10 min, and dried at 37°C for 15 min in a CO₂ incubator. One hour before plating the cells, the coverslips were quenched with RPMI-1640 containing 10% FBS at 37°C. The cells were plated on FITC-gelatin-coated coverslips and cultured in RPMI-1640 for 24 h to quantify formation of invadopodia. Images were visualized by confocal microscopy (Olympus FV1200; magnification of ×200).

After 48 h of treatment with specific drugs or combinations of drugs, the cells were washed and incubated in serum-free medium for 24 h. MMP2 activity was measured by gelatin zymography. Samples were electrophoresed using 10% SDS-PAGE containing gelatin. After electrophoresis, the gel was washed four times with washing buffer [50 mM Tris-HCl (pH 7.5), 100 mM NaCl, and 2.5% Triton X-100], followed by a brief rinse in washing buffer without Triton X-100. The gel was incubated with incubation buffer [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10 mM CaCl₂, 0.02% NaN₃, and 1 mM ZnCl₂] at 37°C for 20 h. After incubation, the gel was stained by Coomassie Blue and destained. A clear zone of gelatin digestion in the gel indicated the presence of MMP2 (55 kDa).

Non-obese diabetic/severe combined immunodeficiency (NOD/SCID) female mice (4 weeks old; average weight, 15 g) were obtained from the Experimental Animal Center of Hubei Province (Animal Study Permit no. SCXK 2013-0004). The mice were maintained under specific pathogen-free conditions (System Barrier Environment no. 00127070) in the Experimental Animal Center of Tongji Hospital of Huazhong University of Science and Technology. All experiments were carried out according to the regulations specified by the Ethics Committee for Tongji Hospital of Huazhong University of Science and Technology. Mice were first intraperitoneally anesthetized with 1% sodium pentobarbital (50 mg/kg of body weight, Sigma-Aldrich, Merck KGaA) and received ovariectomy to eliminate the effects of endogenous estrogen. Next, A549 cells (5×10⁶/100 µl) suspended in PBS were injected into the 4-week-old female NOD/SCID mice via the tail vein. Seven days after injection of the cells, the mice were randomly divided into six groups (n=5/group): mice exposed to E2 (0.09 mg/kg), mice administered E2+Ful (1.46 mg/kg), mice treated with Ful alone, mice exposed to LPS (10 mg/kg), mice exposed to a combination of E2+LPS, and negative control mice. The above-mentioned agents and combinations of agents were administered by subcutaneous injection twice per week for 10 weeks.

At week 10 after exposure, the mice were humanely euthanized by continuous inhalation with 30% CO₂ for 5 min, and were observed for 5 min to ensure the vital activity stopped. All of the experimental mice were sacrificed at the humane endpoint 10 weeks, to ensure the tumor progression time of each group was consistent. After that the lungs were surgically removed. The total number of lung nodules, lung wet weights, and lung cancer metastatic indices were statistically analyzed as previously described (6). The carcinoma tissues were then excised from the lungs; 10 mg of carcinoma tissue per mouse was homogenized in 200 ml radioimmunoprecipitation assay buffer A and subjected to western blotting.

The prognostic value of ERβ/TLR4 was analyzed using the web-based Kaplan-Meier plotter (http://www.kmplot.com/lung), which is a meta-analysis tool for examining gene expression and survival data of 2,437 patients with lung cancer (2018 version) using multiple microarray data.

All experiments were repeated at least twice, and each procedure was performed in triplicate. All data were examined at least three times. Quantitative data are expressed as means ± SD. Statistical significance was established using the SPSS 19.0 statistical software package (SPSS, Inc.). A two-tailed P-value <0.05 was considered statistically significant.

Our previous research and recent studies revealed that ERβ and TLR4 are overexpressed in the cytoplasm and nuclei of NSCLC tumor tissues (15,21,22). To identify the potential roles of ERβ and TLR4 in the development and progression of NSCLC, we collected NSCLC samples from 241 patients with NSCLC and evaluated the expression of ERβ and TLR4 using immunohistochemistry. Our results showed that expression of ERβ was negative in adjacent non-tumor lung tissues, but positive in tumor tissue, and that TLR4 was highly expressed in NSCLC and normal lung tissues (Fig. 1A). Analysis of clinicopathological features indicated that high expression of TLR4 was significantly correlated with lymph node metastasis (χ²=5.732, P=0.017) and adenocarcinoma histological type (χ²=10.152, P=0.001; Table I). However, other clinicopathological features, including sex, age, smoking status, tumor stage, distant metastasis, and pathological tumor-node-metastasis stage, were not directly associated with the expression of ERβ or TLR4 (Table I).

We next used immunostaining to examine the association between expression of ERβ and that of TLR4 in adjacent non-tumor tissues, and in primary NSCLC tissues and metastatic lymph nodes. The results showed the correlation between ERβ and TLR4 co-expression has an increasing Spearman's rank correlation coefficient (Spearman's ρ for short), which showed a significant positive correlation with cancer metastasis stage (adjacent non-tumor lymph nodes, primary tumor, and metastatic lymph nodes). These results are shown in Fig. 1B. The percentage of ERβ-positive stained cells was highly correlated with the level of TLR4 overexpression in metastatic lymph nodes (Spearman's correlation coefficient, r_s=0.411, P=0.004; Table II and Fig. 1E). However, no significant correlation was observed between the expression levels of ERβ and TLR4 in non-tumor lung tissue (r_s=0.069, P=0.468; Table II and Fig. 1C). A slightly stronger correlation between the expression levels of ERβ and TLR4 was observed in primary tumor tissue (r_s=0.344, P<0.001; Table II and Fig 1D). Detailed IHC scores for ERβ and TLR4 expression in primary NSCLC tumors and metastatic lymph nodes and in adjacent non-tumor tissue are shown in Fig. S1A and Table SI. Altogether, these data indicate a positive correlation between protein expression levels of ERβ and TLR4 in clinical NSCLC tissues.

LPS is a significant component of the outer membranes of gram-negative bacteria; LPS triggers TLR4 signaling (13,26). However, it remains unclear whether and how estrogen activates the signaling of ER, TLR4, and its downstream myd88/NF-κB/MMP2 axis. To evaluate the influence of estrogen on the progression of NSCLC, we first exposed NSCLC cell lines to a gradient dose (at 0, 0.1, 1, 10, 100 and 1,000 nM) and then compared the results with those of another group treated with 100 mM E2 at successive time intervals of 0, 1, 6, 12, 24 and 48 h. The results showed that TLR4 was markedly increased after E2 treatment in a dose- and time-dependent manner (Fig. 2A and B; Fig. S2A and B), and ERβ and invasiveness-associated MMP2 were also increased. We then investigated the effects of different treatments on A549 cells by administering DMSO (vehicle/negative control), E2, E2+Ful, or Ful. Our results indicated that the expression levels of TLR4 and MMP2 were significantly increased in the E2-treated group, reduced after treatment with E2+Ful, and reduced after treatment with Ful alone (P<0.05) (Fig. 2D and Fig. S2D). The baseline expression status of ERβ, MMP2, and TLR4 in different cell lines is shown in Fig. 2C and Fig. S2C. As shown in Fig. 2D and Fig. S2D, western blotting analysis further demonstrated an increase in the protein expression of myd88 and phosphorylated p-p65NF-κB, while the total levels of p65NF-κB remained unchanged in the A549 cells. To support these findings, we evaluated the expression of ERβ and TLR4 in A549 and H1793 cells using immunofluorescence analysis. TLR4 overexpression was associated with ERβ activation following exposure to E2 (Fig. 2E and F). Expression of TLR4 and the activation of its signaling pathways were markedly affected by E2. These results indicate a positive correlation between the expression of ERβ and that of TLR4 during NSCLC metastasis.

To investigate how ERβ expression affects the expression of TLR4, ERβ-targeting siRNA or NC siRNA was transfected into NSCLC A549 and H1793 cells to generate a specific ERβ-knockdown cell model. In contrast, an ERβ-overexpression cell model was generated by transfecting an ERβ expressing pcDNA-plasmid into the A549 and H1793 cells. The changes in ERβ and TLR4 expression were evaluated by western blot and immunofluorescence analyses (Fig. 3A and B; Fig. S3A). Our results revealed that removal of endogenous ERβ expression significantly suppressed the expression of TLR4, while overexpression of ERβ significantly promoted the expression of TLR4. In addition, inhibition of ERβ expression significantly decreased expression of MMP2 and myd88. These results indicate that the expression levels of ERβ and TLR4 were positively associated with NSCLC progression.

After treatment with estrogen, cytoplasmic kinase signaling is rapidly activated via ERβ signaling in seconds to minutes. This rapid signaling is termed non-genomic and occurs via non-nuclear ERs located in the membrane or cytoplasm of the cell (27). Our previous research showed that MMP2 upregulation is affected by ERK/P38MAPK and PI3K/AKT signaling pathways activated by estrogen (6). Therefore, we next investigated how the expression of ERβ affects that of TLR4 and downstream signaling pathways. For this, several inhibitors were used to determine the effect of estrogen during blockade of the expression of TLR4/myd88, ERK/P38MAPK, or PI3K/AKT. TAK-242 is a selective inhibitor that suppresses the interaction between TLR4 and its adaptor molecules; this suppression occurs via the intracellular Cys747 residue of TLR4 (23,28). Our results showed that the upregulated expression of TLR4 and MMP2 induced by treatment with estrogen was reversed by treatment with TAK-242 (Figs. 3F and S3E) or with TLR4 siRNA (Figs. 3G and S3F). Next, to support these findings, we performed Transwell migration/invasion assays. Our results demonstrated that invasiveness and aggressiveness of the cells, induced by treatment with E2, were impaired by blocking (Figs. 3D and S3C) or silencing (Figs. 3E and S3D) TLR4 signaling. We then used the pAkt inhibitor LY294002 and the pERK inhibitor PD098059 to treat A549 cells and found that TLR4 expression was significantly abrogated when expression of ERK/P38MAPK and PI3K/AKT was suppressed (P<0.05); this effect was sustained even after exposure to E2 (Figs. 3C and S3B). Altogether, these findings confirm that promotion of NSCLC cell metastasis induced by E2 may be reversed by inhibiting the signaling of TLR4. Our results also showed that activation of the ERK/P38MAPK and PI3K/AKT pathways was required for ERβ-mediated TLR4 upregulation and was necessary to enhance migration and invasiveness of NSCLC cells.

We next used immunofluorescence analysis to evaluate the potential molecular mechanisms underlying the positive association between the expression of ERβ and that of TLR4. For this, A549 cells were treated with 100 nM E2 for 48 h. Indirect immunofluorescence analysis indicated that endogenous expression of ERβ and TLR4 was primarily co-localized in the cytoplasm, while the nuclear region showed sparse co-expression (Fig. 4A). Similar results were observed in the cytoplasm of another NSCLC cell line, H1793. In this cell line, co-localization was particularly visible under higher magnification and at a higher resolution. These data indicate that ERβ and TLR4 proteins likely form a complex or interact with each other.

Previous studies have shown that the amino-terminal region of ER enables it to dimerize or bind to co-activators, co-repressors, regulators, and ligands. This binding is responsible for the promotion of enhanced tumorigenesis (27,29). Our previous findings indicated a positive signaling association and interaction between ERβ and TLR4. This prompted us to examine whether there are direct physical and molecular interactions between ERβ and TLR4. To determine whether endogenous ERβ and TLR4 interact with each other, we co-immunoprecipitated ERβ with TLR4 using cell lysates isolated from A549 cells. TLR4 was readily detected in ERβ immunoprecipitates (Fig. 4B), while ERβ was found in TLR4 immunoprecipitates (Fig. 4C). However, the interaction of endogenous myd88 or MMP2 with ERβ was not detected (Fig. 4B). Hence, our results indicate that there is an interaction between ERβ and TLR4. These data suggest that the interaction of ERβ and TLR4 in cellular cytoplasm contributes to NSCLC metastasis.

After confirming that exposure to E2 induced endogenous interaction of ERβ and TLR4 and increased TLR4 protein levels, we assessed the extent to which the metastatic aggressiveness of the human NSCLC A549 and H1793 cell lines was mediated by exposure to E2, LPS, and combined administration of E2+LPS. The wound-healing assay revealed that cells treated with E2+LPS significantly increased their rate of lateral migration into wounds introduced on confluent cellular monolayer, compared with the rate of lateral migration shown by the single-drug-treated groups and inhibitor-treated groups (A549 cells: Fig. 5A and B, P<0.05; H1793 cells: Fig. S4A and B, P<0.05). To detect and quantify the aggressiveness of NSCLC cells, we next performed Transwell migration\invasion assays. The highest number of invading cells was found in the E2+LPS treated group. These results indicate that treatment with E2+LPS increased the metastatic abilities of the NSCLC cells (A549 cells: Fig. 5C and D, P<0.05; H1793 cells: Fig. S4C and D, P<0.05). Western blotting and immunofluorescence staining were used to determine the expression levels of different proteins involved in the ERβ/TLR4 interaction and downstream signaling pathways (A549 cells: Figs. 5E and F and S4E; H1793 cells: Figs. 5G and S5C and D). The results of the densitometric analysis showed that the levels of ERβ, TLR4, myd88, and MMP2 proteins and phosphorylation of p65NF-κB (p-p65NF-κB) induced by the treatment with E2+LPS were significantly higher than were those induced by treatment with E2 or LPS. Taken together, our findings indicate that activation of ERβ and TLR4 synergistically promoted the metastatic abilities of the NSCLC cells.

Increased cell invasion in metastatic progression is induced by invadopodium formation followed by extracellular matrix (ECM) degradation (30,31). Hence, we explored whether the combined exposure to E2 and LPS accelerates invadopodium formation and ECM degradation. We performed in vitro Matrigel 3D spheroid invasion assays using A549 and PC9 cells administered E2, LPS, E2+LPS, and DMSO (vehicle) to evaluate invasiveness in the cell lines. Density and length of cell-invasion structures were evaluated on days 1, 3 and 6. Compared with negative control treatment, exposure to E2 or LPS alone significantly increased the invasive capability of cells, which formed invadopodia on day 6. The group treated with E2+LPS showed accelerated formation of invadopodia on day 3 (A549 cells: Fig. 6A and B). The H1793 cells failed to form spheroids in Matrigel media; therefore, we used PC9 NSCLC cells for this procedure (PC9 cells: Fig. S5A and B). Invadopodia appear as accumulations of F-actin associated with dark areas of fluorescent gelatin degradation (31,32). To fully determine whether the combination of E2 and LPS enhanced the function of invadopodia, we conducted a fluorescent gelatin degradation assay and gelatin zymography analysis. The co-localization of F-actin with gelatin degradation induced in cells treated with estrogen or LPS related-drugs compared with that in unstimulated cells is shown in Fig. 6C. A549 cells treated with E2+LPS showed over 400 µm² dark degradation area/cell, while single-drug-treated groups of cells showed barely 200 µm² degraded area/cell (Fig. 6C and D), and we repeated this procedure with H1793 cells (Fig. S6A and B). Consistently, the results of gelatin zymography also showed that exposure to E2+LPS induced more ECM degradation in the cells than did exposure to the single agents (Fig. 6E and F). Together, these data indicate that the combination of estrogen and LPS strongly accelerated and enhanced invadopodia function in NSCLC cells.

The effect of combined exposure to E2 and LPS in cancer metastasis, which was demonstrated in vitro, was further confirmed in an animal model. For this, female ovariectomized NOD/SCID mice were injected with A549 cells into the tail vein. After 65 days of treatment, mice from each group (n=5) were sacrificed, and the number of metastatic nodules in the lungs was examined at the indicated time points. As expected, mice exposed to the combination of E2 and LPS showed higher lung wet weights, numbers of lung metastasis lesions, and metastatic indices than did mice injected with normal saline (negative control) or other treatments (Fig. 7A, and C-E). Hematoxylin and eosin staining for metastatic nodules is shown in Fig. 7F. We then used western blotting and immunohistochemistry to assess protein levels in each group of mice. As shown in Fig. 7G and H, the expression of ERβ and TLR4 in the E2+LPS-treated group was significantly higher than that in groups exposed to either of E2 or LPS alone, and expression levels were decreased in the E2+Ful- and Ful-treated groups. Consistently, invasiveness-associated overexpression of MMP2 was higher in the group treated with E2+LPS than that in the groups subjected to other treatments. Additionally, the immunohistochemical staining of ERβ and TLR4 in murine metastatic lung tissues is shown in Fig. 7B. Altogether, our findings showed that the combination of E2+LPS synergistically increased lung metastasis in the mouse tumor model.