The free survival of lung cancer patients with different LOX gene expression was analyzed using Kaplan-Meier Plotter². The expression of EGFR and LOX in human tumor tissue was analyzed by the Oncomine^TM database³. The correlation of EGFR and LOX gene expression in human NSCLC tissue was performed by TCGA Research Network⁴.

Human lung cancer cell lines, NCI-H1975, HCC827, A549, H460, and NCI-H1299 were obtained from the Cell Bank of the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China). The cells were cultured in RPMI-1640 (Gibco) with 10% Foetal Bovine Serum (HyClone). All the lung cancer cells were recently authenticated by short tandem repeat (Genetic Testing Biotechnology Corporation, Suzhou) and were tested for mycoplasma contamination.

Human lung cancer specimens were collected from the National Engineering Center for Biochip at Shanghai (SBC), China. The study was approved by the Research Ethics Committee in Taizhou Hospital of Zhejiang Province and written informed consent was obtained from all participants. In total, 30 patients who underwent surgery for histologically proven lung adenocarcinoma were selected in this research. There were a total of 14 men and 16 women, whose ages range from 37 to 71 years.

The antibodies and reagents included EGFR (4267s, CST), P-EGFR (Y1068) (3777S, CST), LOX (A10960, ABclonal), AKT (9272S, CST), P-AKT (S473) (#4060, CST), ERK1/2 (4695S, CST), P-ERK1/2 (Thr202/Tyr204) (4370s, CST), P38 MAPK (8690S, CST), P-P38 MAPK (T180-Y182) (4511S, CST), SAPK/JNK(56G8) (9258S, CST), P-SAPK/JNK (T183/Y185) (4668S, CST), LY294002 (S1105, Selleck), U0126 (S1102, Selleck), SB203580 (S1076, Selleck), SP600125 (S1460, Selleck), and TGF-α(Pepro, Tech).

The total cell lysates were extracted with lysis buffer containing protease and phosphatase inhibitors. The proteins were fractionated by 6–15% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and were transferred to PVDF membrane (Millipore, United States). The membranes were blocked with 5% BSA-TBST for 1 h at room temperature and were incubated with primary antibodies (diluted in 5% BSA-TBST) for 18 h at 4°C. Next, they were probed with secondary antibodies for 1 h at room temperature. The expression of the target proteins was detected by the Immobilon Western chemiluminescent HRP Substrate (Millipore, United States). Quantification of protein bands was carried out using densitometry and the respective densitometry graphs are shown in the Supplementary Figures.

Total cellular RNA was isolated with the TRIzol^® Reagent (Vazyme) and was reverse transcribed with the HiScript^TM QRT SuperMix for qPCR (Vazyme). The mRNA level was measured with the SYBR Green master mix (Vazyme). The amount of mRNA for each gene was standardized with the internal control (18s mRNA). Each treatment group was compared with the control group to show the relative mRNA level.

All the siRNA were synthesized by GenePharma. The cells were transiently transfected using Lipofectamine^TM 2000 (Invitrogen) according to the manufacturer’s recommendations.

The cell proliferation assays were performed by MTT. The cells were seeded into 96-well plates in 180 μL of medium per well. After 12 h, the cells were treated with silibinin at specified concentrations for additional 12, 24, 48, or 72 h. The cells were treated with 5 mg/mL of MTT solution (20 μL/well) for 4 h. Then, the supernatant was replaced and DMSO (150 μL/well) was added. After the formazan was dissolved, the absorbance was measured at 570 nm using a Universal Microplate Reader (EL800; Bio-tek Instruments Inc.). The inhibitory ratio was calculated by the following formula: inhibitory ratio (%) = (1–average absorbance of treated group/average absorbance of control group) × 100. The results are presented as mean standard deviation (SD). In addition, triplicate experiments were performed in parallel.

The migration assay was detected by 24 Well Transwell Chambers (Millipore, United States). The cells were seeded at a density of 6–8 × 10⁴ cells per chamber with drugs at specified concentrations both in the lower and the upper chamber. The cells were allowed to migrate for 16 h. After discarding the medium, the migrating cells were stained with DiffQuik Stain Set (Jiancheng Bioengineering Institute), photographed, and counted with Image-Pro Plus software.

Wound healing assay was carried out to determine the migration ability of the tumor cells. The cells were seeded into 6-well plates and incubated in complete medium to 90% confluence. A sterilized pipette tip was used to generate wounding across the cell monolayer, and then, the cells were washed twice with PBS, replaced with fresh media and treated with drugs at specified concentrations for another 12 h. The cells migrated into the wounded area were visualized and photographed under the inverted microscope.

Schrödinger was used for the molecular modeling studies. The crystal structure of the EGFR kinase domain (PDB ID: 3W2S) was prepared using Protein Preparation Wizard. The structure of silibinin (ZINC02033589) was prepared using LigPrep. The calculation was performed based on the force field OPLS (optimized potentials for liquid simulations) 2005 choosing water as the solvent. The following structure was obtained from the result of 1000 calculation cycles.

Five to six-week-old female BALB/c mice (18 to 22 g) were supplied by the Model Animal Research Center of Nanjing University. NCI-H1975 cells were collected in serum-free medium and were mixed with same volume Matrigel (Coring). The mice were anesthetized by isoflurane inhalation, and then, 50 μL of cell suspensions (1 × 10⁶) were orthotopically injected through the intercostal space into the lung immediately after making a small skin incision. The mice were randomly divided into four groups (6 mice/group) 11 days after the injection. Silibinin (70 mg/kg, 140 mg/kg) and WZ4002 (25 mg/kg) were administered intragastrically in the treatment groups daily for 3 weeks, while the control animals were administered the same vehicle. After 3 weeks, the metastatic nodes were detected by gross examination. The study was conducted in accordance with the standards established by the Experimental Animal Care Commission in China Pharmaceutical University.

The P-EGFR and LOX levels in the lung tumors were evaluated by IHC using P-EGFR and LOX antibodies. The tumor tissues from the nude mice were formalin fixed and paraffin embedded. The paraffin embedded tissues were de-paraffinized, rehydrated, rinsed and immersed using 10 mM sodium citrate (pH 6.0). After treatment with methanol containing 3% hydrogen peroxide, the tissues were staining with P-EGFR, LOX, and secondary antibodies.

The relationship between EGFR and LOX expression was analyzed via the Spearman rank correlation test. The results were presented as the mean ± SD from triplicate experiments performed in a parallel manner unless otherwise indicated. Statistically significant differences were determined using GraphPad Prism 6 software. A value of P < 0.05 was considered significant.

To test the relationship between LOX expression and the prognosis of cancer patients, we used Kaplan–Meier plotter⁵ to analyze the correlation.

The results showed that a high expression of LOX was closely related to the poor prognosis of lung cancer patients (Figure 1A), and was more associated with lung adenocarcinoma patients but not lung squamous cell carcinoma patients (Figure 1B). In further exploration, data showed that influence of LOX expression on the survival of lung adenocarcinoma patients is independent of gender and smoking history (Figures 1C,D). As is reported, half of the lung adenocarcinoma patients were accompanied with an EGFR mutation (Midha et al., 2015). Thus, we suspected a relationship between EGFR and LOX expression. Firstly, according to the Oncomine^TM database⁶, the expression levels of EGFR and LOX were higher in lung adenocarcinoma tissues (Figures 2A,B). Furthermore, TCGA Research Network⁷ also indicated that there might be a correlation between EGFR and LOX mRNA expression in lung adenocarcinoma patients (Figure 2C). In fact, both the overexpression and mutation of EGFR eventually led to the phosphorylation and concomitant activation of downstream signaling, promoting the proliferation, metastasis, and resistance to chemotherapy in cancer cells. Therefore, we tested the expression of P-EGFR and LOX in five NSCLC cell lines, the data showed that expression of LOX was relatively higher in the cell lines with high expression of p-EGFR (NCI-H1975 and HCC827) than that with low P-EGFR expression (H460, NCI-H1299, and A549) (Figure 2D and Supplementary Figure S1). Moreover, human tumor tissue microarray further confirmed the possible correlation between P-EGFR and LOX (Figures 2E,F). Conclusively, these data suggest that there maybe a regulation relationship between P-EGFR and LOX expression.

To further explore the relationship, we inhibited EGFR by WZ4002 in NCI-H1975 and HCC827 cells. Interestingly, the expression of LOX was also dramatically inhibited (Figures 3A,B and Supplementary Figures S2A,B). Meanwhile, when EGFR was knocked down in two NSCLC cells, the expression of LOX was remarkably decreased at both the mRNA and protein levels (Figures 3C–H and Supplementary Figures S2C,D). In addition, the phosphorylation of EGFR and LOX expression in EGFR wild type NSCLC cell line A549 were both up-regulated when treated with the EGFR agonist TGF-α. (Figure 3I and Supplementary Figure S2E). Here, we concluded that LOX expression is regulated by the phosphorylation of EGFR in NSCLC cell lines, but the mechanism needs to be further explored.

To evaluate the regulatory mechanism by which LOX is regulated by EGFR, we focus on the four classical EGFR signal pathways (PI3K/AKT, MEK/ERK, P38 MAPK, and SAPK/JNK) with specific inhibitors (LY294002 for PI3K, U0126 for MEK, SB203580 for P-38, and SP600125 for JNK) (Keld and Ang, 2011). As shown in Figures 4A,C (Supplementary Figures S3A–D), LOX expression was dramatically down-regulated by PI3K/AKT and MEK/ERK inhibition at both the mRNA and protein levels in the NCI-H1975 cell, and the inhibition of the SAPK/JNK pathway also leads to the decreased expression of LOX protein. Moreover, when treated with the PI3K/AKT, MEK/ERK, and P38 MAPK inhibitors in another EGFR mutation NSCLC cell line HCC827, LOX expression at the mRNA and protein levels was inhibited. After the inhibition of the SAPK/JNK pathway, we obtained consistent data with the NCI-H1975 cells (Figures 4B,D and Supplementary Figures S3E–H). However, SAPK/JNK inhibition decreased the protein level of LOX but not mRNA level. This may because PI3K/AKT and MEK/ERK pathway regulate LOX expression through transcriptional level, while SAPK/JNK pathway via ubiquitin–proteasome system or lysosomal system. Next, we will investigate the underlying mechanism in the following study. Briefly, as is shown above, LOX expression was mainly regulated by the EGFR signal pathways PI3K/AKT, MEK/ERK, and SAPK/JNK in NSCLC cell lines.

Accumulating evidences show that silibinin, a classical anti-fibrosis drug, exhibits strong anti-cancer activity in several human malignant tumors (Wu et al., 2013; Bosch-Barrera et al., 2016), but the underlying mechanism is still unknown. In consideration of one important molecule during fibrosis is LOX in hepatocyte, we suspected that whether the EGFR/LOX pathway plays a role in its anti-metastasis effect. Above all, we carried out an MTT assay to determine the effect of silibinin on the proliferation of NCI-H1975 cells, the IC₅₀ was 96.56 ± 2.755 μM (Figure 5A). Meanwhile, the effect of silibinin on the migration of NCI-H1975 cells was measured by wound healing and transwell assay. As shown in Figures 5D,E, silibinin significantly inhibited the migration of NCI-H1975 cells but scarcely showed any proliferation inhibition effect at the treatment concentration (Figures 5B,C). Here, the data indicated silibinin significantly inhibited the migration of NCI-H1975 cells.

To explore if the EGFR/LOX signal pathway accounts for the anti-metastasis effect of silibinin, we investigated the relative signal pathways by PCR and Western Blot. The results showed that LOX mRNA expression was significantly decreased after silibinin treatment for 48 h (Figure 6A), silibinin also inhibited the EGFR/PI3K/LOX signaling pathway (Figure 6B and Supplementary Figure S4). Our above study shows PI3K/AKT, MEK/ERK, and SAPK/JNK pathways all participate in the regulation of EGFR to LOX. However, in consideration silibinin is a natural medicine, other signal molecules affected by silibinin may also act on p-JNK and p-ERK pathways. Meanwhile, our treatment dose is far below the doses in other researches (Bhatia and Falzon, 2015; Shukla et al., 2015). Thus, silibinin regulates LOX expression mainly through PI3K/AKT pathway in our treatment doses. When the LOX inhibitor (BAPN) blocked the migration of NCI-H1975 cells, it also showed that pretreatment with the LOX inhibitor attenuated the anti-metastasis effect of silibinin (Figures 6C,D). Though, silibinin is still able to exhibit its anti-metastasis effect maybe via other signal pathways after the inhibition of LOX. But the fold change of silibinin on migration after LOX inhibition is much smaller than it without LOX inhibition.

Meanwhile, cancer cells often change their cellular and molecular morphology during the metastasis process. Then we investigated the expression of epithelial-to-mesenchymal transition (EMT) molecules in NCI-H1975 cells treated with LOX inhibitor. The results showed that LOX inhibition supressed the expression of vimentin, slug, snail and N-cadherin, and increased the expression of epithelial phenotypic markers such as E-cadherin, ZO-1, and β-cantenin (Figure 6E and Supplementary Figures S5A–G), suggesting LOX inhibition could inhibit EMT in NSCLC cells. Coincidently, the previous study reported that silibinin could repress the EMT to suppress the invasive property of metastatic prostate cancer cells (Wu et al., 2010; Cufi et al., 2013). According to our result that silibinin could decrease the expression of LOX (Figures 6A,B), it suggested the inhibition of EMT by silibinin may be via LOX. In addition, the phosphorylation of EGFR was also decreased with the inhibition of LOX, but not the total expression of EGFR (Figure 6F and Supplementary Figures S5H–I), which indicating there may be a regulation loop between P-EGFR and LOX. In conclusion, silibinin exhibited an anti-metastasis effect possibly via LOX regulated EMT molecules. But whether silibinin inhibits migration by targeting EGFR is unknown.

Molecular docking studies predict potential interactions of the proposed protein with the selected molecule, which is a structural modeling approach to study possible binding sites for cancer therapeutics. To explore the involvement of EGFR as a therapeutic target of silibinin, we first used molecular docking methods to predict the binding affinity of EGFR and silibinin. Based on the receptor structure combined with the docking pharmacophore virtual screening method with G-score (-11.069), silibinin formed four hydrophobic interactions (GLY 724, ASP 855, LYS 745, and MET 793) and 1 π–π stacking (PHE 723) with EGFR. This result suggests that silibinin could strongly bind to the functional domains of EGFR and affect the bioactivity of EGFR (Figures 7A–C). Then, as shown in Figure 7C, LOX expression was significantly decreased after treatment with silibinin and EGFR-siRNA. Meanwhile, the fold change of silibinin on migration after knockdown of EGFR is much smaller than it without knockdown of EGFR. These result suggested that knockdown of EGFR partially attenuated the effect of silibinin on LOX expression and migration (Figures 7D,E). In other words, silibinin exhibited anti-metastasis partly via targeting EGFR. Therefore, the data suggested that the EGFR/PI3K/LOX pathway might contribute to the anti-migration effect of silibinin.

The regulatory relationship between the phosphorylation of EGFR and LOX, the inhibitory effect of silibinin on NSCLC cell migration was confirmed in vitro. However, whether silibinin inhibited NSCLC metastasis through EGFR/LOX in vivo remains unknown. In the orthotopic implantation metastasis model, we discovered that EGFR inhibitor WZ4002 and silibinin not only significantly prolonged the survival of the mice but also decreased the formation of metastasis nodes (Figures 8A,C,D). Moreover, after treatment with WZ4002 or silibinin, the body weight of mice was also increased (Figure 8B). Furthermore, the expression of LOX was inhibited, which is consistent with the inhibition of P-EGFR by the treatment with WZ4002 and silibinin in vivo (Figure 8E). In summary, as shown above, it was confirmed that silibinin decreased NSCLC metastasis via the EGFR/LOX pathway in vivo.