A total of 76 cases of LUAD were examined in our present study, which consisted of 16 cases from the Fifth Affiliated Hospital of Harbin Medical University from 2009 to 2014; 18 from Daqing Oilfield General Hospital from 2008 to 2017; 42 from the Daqing Longnan Hospital from 2018 to 2019. Clinicopathological findings including their gender, age, smoking history, histological type, tumor size, tumor differentiation, lymph node status, and clinical stage were all retrieved from the medical records at these three institutions. All the specimens examined were retrieved from surgical pathology files of resected specimens, and all the patients were treatment naïve before the surgery. Hematoxylin and eosin stained tissue slides were all independently evaluated by two of the authors (BZ and LL) blinded to the clinical data of the patients. In addition, 17 cases of non-pathological or non-tumorous lung (NTL) tissues obtained by autopsy were retrieved from the Forensic Center of Harbin Medical University-Daqing as controls.

The research protocol of our present study was approved by the Institutional Review Board of Harbin Medical University-Daqing (Approval number: HMUDQ2020120501). All procedures were conducted in Accordance with the Declaration of Helsinki and International Ethical Guidelines for Biomedical Research Involving Human Subjects (CIOMS).

The human LUAD cell lines (A549, H1975, and H-1299) and the breast carcinoma cell lines (MCF-7 and MD-MBA-231) used in this study were purchased from the Cell Bank of the Chinese Academy of Sciences. Cell line authenticity was verified using STR analysis and mycoplasma was tested by the supplier prior to their shipment. The cells were maintained in DMEM medium (Gibco, Cat# 11995065) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Biological Industries) at 37°C with 5% CO₂. Before E₂ treatment, the medium was changed to complete phenol-red free DMEM (Gibco, Cat# 21063029) supplemented with 5% charcoal stripped FBS (Hyclone, Cat# SH30068.03). 17β-Estradiol (E₂) was purchased from Sigma-Aldrich (Cat# E2257). E₂ was dissolved and stocked in ethanol at the concentration of 1 mM before addition into the culture medium. When being used for the treatment of cells, the final volume ratio of E₂-ethanol solution to the medium was no more than 1:10,000 (V:V). The primary antibodies used in this study were purchased as follows: anti-PELP1, from Bethyl (Cat# IHC-00013); anti-ERα, from Santa Cruz (Cat# sc-543); anti-ERβ, from Abcam (Cat# ab288); anti-PCNA, from Boster (Cat# BM0104); anti-MMP-9, from Abcam (Cat# ab76003); anti-β-Actin, from Santa Cruz (Cat# sc-69879).

Ten percent formalin-fixed and paraffin-embedded tissue blocks of the selected samples were retrieved from the archives for detecting PELP1, ERα, ERβ, PCNA, and MMP-9 protein expression. All samples were sliced to 4-μm thick serial section, deparaffinized with xylene, and rehydrated in alcohol gradients. Endogenous peroxidase was blocked with 3% hydrogen peroxide in methanol for 30 min. Antigen retrieval was performed by microwave (500 W) irradiation for 15 min. Primary antibodies against PELP1 (1/200), ERα (1/200), ERβ (1/100), PCNA (1/100), MMP-9 (1/100) were incubated overnight at 4°C. Real Envision Detection system (DAKO, Denmark, Cat# K5007) was used following the manufacturer’s instruction. Sections were visualized with DAB and counterstained with hematoxylin. Negative controls were performed by omitting the primary antibodies and substituting them with antibody dilution buffer (DAKO, Denmark, Cat# S2022).

The evaluation of immunoreactivity was performed independently by two of the authors (SC and ML). H-Score was used to assess PELP1 immunoreactivity. Briefly, PELP1 immunointensity was tentatively scored as 0, 1, 2, or 3, and the percentage of positive cells was determined for each score to yield a final score in the range of 0–300. The optimized cutoff points in the Habashy et al. [21] study were also employed in this study and the cut-off points were defined using the X-tile program. PELP1 immunoreactivity was tentatively classified into negative (H-score < 5), moderate (5 ≤ H-score < 170) and strong (170 ≤ H-score). When assessing ERα and ERβ immunoreactivity, tumor nuclear immunoreactivity of ≥1% was defined as positive. The evaluation of MMP-9 immunoreactivity followed the method reported previously [31]. When assessing PCNA immunoreactivity, the method of Grossi et al. was used [32].

Three siRNA sequences and nonspecific control (NC) siRNA targeting the PELP1 gene were designed and synthesized by Jima Pharmaceutical Technology (Shanghai, China). The cell transfections were performed using X-tremeGENE siRNA Transfection Reagent (Roche, Cat# 04-476-093-001) according to the manufacturers’ protocol.

A549, H1975, and H-1299 cells were seeded in 96-well plates at a density of 1 × 10⁴ cells per well in complete phenol-red free DMEM medium with 5% charcoal stripped FBS under standard conditions (37°C and 5% CO₂). After treatment with different treatment factors, a total of 20 μl MTT (5 mg/ml) was added to each well and incubated for additional 4 h, then the supernatant was aspirated and formazan crystals were dissolved in 0.5 M dimethylformamide and 20% sodium dodecylsulfate (SDS). Optical density (OD) was recorded at 490 nm using a microplate luminometer.

A549 cells were seeded into six-centimeter plates in triplicates at a density of 600 cells/plate in complete phenol-red free DMEM medium and incubated for 24 h, then transfected with PELP1-siRNA-2 (siR-2) or NC for 24 h. The cells were then incubated with or without E₂ for 2 weeks under standard conditions (37°C, 5% CO₂ and 5% charcoal stripped FBS) and the culture medium with or without E₂ was refreshed every 48 h. For counting the colonies, the cells were washed twice with phosphate-buffered saline (PBS), fixed with 5 ml methanol for 15 min and then stained with Giemsa for 30 min. Colonies containing more than 50 cells were counted under light microscopy with 100× magnification. The colony forming rate was calculated by dividing the number of positive colonies by the total number of cells seeded.

Proteins were extracted with RIPA lysis buffer. Total 40 μg of protein for each sample was subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto a Nitrocellulose blotting membrane. The membrane was blocked in 5% skimmed milk, then washed three times with Tris buffered saline with 0.05% Tween 20 (TBS-T), and probed with PELP1 (1/500), ERα (1/1000), ERβ (1/500), PCNA (1/200), MMP-9 (1/2000) or β-Actin (1/5000) primary antibodies overnight at 4°C. The samples were incubated with the appropriate secondary antibodies for 1 h. Quantification of the bands was analyzed with the NIH ImageJ program.

After treatment, the cells were re-suspended with 70% ethanol in a final concentration of 10⁶ cells/ml, and then incubated with propidium iodide (PI) at 4°C overnight. The cells in different cell cycle phases were counted on a FacScan (Becton Dickenson) with CellQuest software (Becton Dickenson) and analyzed using Cyflogic software (CyFlo Ltd.). The proportion of cells in each cell cycle phase was determined and the data were presented from three independent experiments.

Cells were cultured in a monolayer reaching 70% confluence on six-centimeter plates. Monolayers were wounded by scratching with a 100-μl pipette tip and subsequently washed three times with PBS. During the treatment with E₂ and/or siR-2, the images of the wound site were captured using a light microscopy (Ti, Nikon Instruments Inc., Japan) (magnification 200×) at the same location, respectively at 0 and 24 h. The cell migrated areas were then measured using NIH ImageJ program.

Matrigel invasion assay was performed using 24-well transwell plates (Corning, Cat# 3422) with polycarbonate filters (pore size, 8 μm). Attenuated matrigel was made by diluting matrigel to 1 mg/ml with serum-free medium on ice (4°C). 100 μl of attenuated matrigel was added to the upper chamber of the transwell plates. Next, matrigel was rinsed using serum-free medium and 100 μl cell suspension (2 × 10⁵ cells/ml) was added to the upper chamber. For the lower chamber, 600 μl medium with 20% FBS was added. After 20–24 h of incubation in 37°C incubator, the transwell was washed with PBS twice and then the cells were fixed using 5% glutaraldehyde. Following fixation, the cells were rinsed twice in PBS and Giemsa staining was used to identify those invaded. The transwell was cultured at room temperature for 0.5 h and then washed in PBS twice. Finally, the invasion of cells in the transwell chambers was detected under light microscopy (90i, Nikon Instruments Inc., Japan) and invasive cells were calculated using the NIH ImageJ program.

Correlation between overall survival (OS) of LUAD patients and PELP1 expression (Affymetrix ID: 215354_s_at) was analyzed by the online cancer survival analysis tool Kaplan-Meier Plotter (http://kmplot.com). “Only JetSet best probe set” and “Auto select best cutoff” were selected in the analysis; Array quality control was set as “exclude biased array”. A total of 719 LUAD and 524 LUSC samples were included in the analysis. Kaplan-Meier analysis was performed to determine the correlation between the expression of PELP1 and OS of LUAD and LUSC patients.

Statistical analysis was performed with SPSS 17.0 statistical software (Chicago, United States). Association between PELP1 expression and clinicopathological variables in tissues was studied with Chi-square Test (Fishers Exact Test) or Spearman correlation test; ANOVA was used for analyzing results from in vitro assays (MTT, colony formation assay, western blot, flow cytometry, wound healing assay and transwell assay), and LSD method was chosen for Post Hoc Multiple Comparison. A p value < 0.05 was considered significant.

PELP1 immunoreactivity in 76 cases of LUAD and 17 cases of non-tumor lung (NTL) tissues were examined. In NTL tissues, PELP1 was immunohistochemically negative or moderate immunoreactivity in the nuclei of few lung epithelial cells and scattered lymphocytes. In LUAD tissues, PELP1 positive immunoreactivity was mainly located in the nuclei of tumor cells. In some cases, PELP1 moderate immunoreactivity could also be detected in some stromal fibroblast cells or lymphocytes. No cytoplasmic or extracellular immunostaining of PELP1 was detected in NTL or LUAD tissues (Figure 1).

Among the NTL cohort, 8/17 cases were immunohistochemically negative for PELP1, 9/17 moderately positive for PELP1 but none markedly positive. Among the LUAD cohort, 13/76 were negative for PELP1, 45/76 moderately positive, and 18/76 markedly positive. PELP1 status in LUAD was significantly higher than that in NTL (χ² = 9.483, p = 0.006).

ERα, ERβ, PCNA, and MMP-9 were immunolocalized in the 76 cases of LUAD. Positive ERα immunoreactivity in LUAD cases was mainly located in the nuclei of tumor cells with its weak cytoplasmic immunoreactivity (Figure 2A) but ERβ immunoreactivity was only detected in the tumor nuclei without any cytoplasmic localization (Figure 2B). PCNA was exclusively immunolocalized in the nucleus (Figure 2C) and MMP-9 mainly distributed in the cytoplasm of LUAD cells (Figure 2D).

The status of PELP1 was correlated with clinicopathological variables of the patients including the status of ERα, ERβ, PCNA, and MMP-9 in LUAD. Significantly higher PELP1 was associated with tumors of higher grade with regard to the tumor differentiation, lymph node metastasis and clinical stage of the patients. In addition, PELP1 was significantly positively correlated with ERα, ERβ, and PCNA in LUAD patients but no significant association with other clinicopathological variables including patient’s gender, age, smoking history, tumor size as well as the expression of MMP-9 (Table 1).

In order to further study the functional roles of PELP1 in estrogen regulated proliferation and migration of LUAD cells, two LUAD cell lines, A549 and H-1299 were tentatively selected for the in vitro study in our present analysis. Western blot was first performed to examine PELP1, ERα, ERβ expression in MCF-7 (ERα, ERβ positive and PELP1 highly expressed) and MDA-MB-231 (ERα, ERβ negative and PELP1 highly expressed) breast carcinoma cells as control. A549 cells demonstrated high expression of PELP1 protein, nearly equivalent to that in MCF-7 and MDA-MB-231 cells but the protein level of PELP1 in H-1299 cells was significantly lower than that in MCF-7 and MDA-MB-231. A549 demonstrated weak ERα and robust ERβ expression, while in H-1299 cells, both ERα and ERβ were very low (Figure 3A).

MTT and wound healing assay were then performed to further explore the roles of estrogen on the proliferation and migration of two LUAD cells. Following the treatment with different concentrations of E₂ for 24 h, A549 demonstrated an E₂ concentration-dependent increment of proliferation and migration reaching their maximal effects, respectively at the concentration of 10 and 100 nM of E₂ (Figures 3B,E). However, these pro-proliferation and pro-migration actions of E₂ were not detected in H-1299 cells (Figures 3C,F). Treated A549 combining with 5uM Fulvestrant (Ful) significantly reduced the pro-proliferation and pro-migration activities of E₂ on the cells (Figures 3D,G). The expression of PELP1 in A549 and H-1299 were examined by western blot after the cells were treated with E₂ (10 nM) in combination with or without Ful (5 nM) for 48 h. In A549, the protein levels of PELP1were significantly increased after E₂ treatment and inhibited by Ful treatment, but either E₂ alone or in combination with Ful showed no influence on the PELP1 expression in H-1299 (Figure 3H). In addition, the pro-proliferation and pro-migration activities of E₂ were also detected in ERs, PELP1 double positive H1975 LUAD cell line (Supplementary Figure S1). These results all indicated that both ERs and PELP1 were necessary for E₂ to exert its pro-proliferation and pro-migration actions on LUAD cells.

In order to further identify the roles of PELP1 in the proliferation, migration, and invasion of A549 cells induced by E₂, three siRNA sequences were designed targeting PELP1 in A549 cells. Results of western blot analysis indicated the three PELP1-siRNAs successful knocking down PELP1 and did by no means influence the expression of ERα or ERβ (Figure 4A). siR-2 demonstrated the highest knock-down efficiency and was therefore selected for subsequent analysis. MTT, western blot, and colony formation assay were all employed to assess the influence of knocking down PELP1on E₂ induced proliferation of A549 cells. Under the stimulation of 10 nM E₂, siR-2 transfected cells demonstrated significantly reduced cell viability compared with NC transfected cells in MTT assay (Figure 4B). Western blot assay also demonstrated that siR-2 prevented the increased PCNA protein level stimulated by E₂ in A549 cells (Figure 4C). In addition, we also examined the cell cycle progress under the stimulation of E₂ in NC or siR-2 transfected A549 cells. In the NC transfected A549 cells, E₂ significantly increased the proportion of S and G2/M stage cells. However, in siR-2 transfected cells, E₂ did not show significant influence on the cell cycle progress (Figure 4D). In addition, colony formation assay revealed knock-down of PELP1 using siR-2 significantly inhibited colony formation induced by E₂ in A549 cells (Figure 4E). Both wound healing and transwell assays demonstrated that knock down of PELP1 significantly inhibited E₂ (100 nM) induced migration and invasion of A549 cells (Figures 4F,G), and western blot also revealed that knock-down of PELP1 attenuated E₂ induced MMP-9 expression (Figure 4H). Knocking down PELP1 in H1975 also attenuated E₂ induced proliferation and migration as well as the expression of PCNA and MMP-9 of the cells (Supplementary Figure S2). Those results above all confirmed a pivotal role of PELP1 in E₂ induced proliferation, migration, and invasion of LUAD cells.

In order to explore the potential prognostic value of PELP1 expression on LUAD patients, we performed survival analysis using the online survival analysis tool Kaplan-Meier Plotter. The Gene expression data and survival information utilized in the tool were based on the database of GEO, EGA, and TCGA. This in silico analysis demonstrated LUAD patients with high PELP1 expression had significantly shorter OS than those with low PELP1 expression (p = 0.002 for 60 months following-up, and p = 0.006 for 120 months following-up). However, this prognostic impact of PELP1 was not detected in the patients with LUSC (p = 0.12 for 60 months following up, and p = 0.19 for 120 months following up) (Figure 5).