Tissue specimens were obtained from the Department of Pathology, The Third Affiliated Hospital of Kunming Medical University (Tumor Hospital of Yunnan Province) between May 2003 and July 2010. Specimens were fixed with 10% neutral formalin for 72 h at room temperature and were then embedded in paraffin. The specimens consisted of 52 primary NSCLC tumor tissues and 49 benign lung tissues. The NSCLC specimens were obtained from 52 patients: 40 with adenocarcinoma and 12 with squamous cell carcinoma, including 30 men and 22 women, with ages ranging between 34 and 70 years (mean age, 59 years). The benign lung tissues were obtained from 49 patients with benign pulmonary diseases: 29 men and 20 women, with ages ranging between 32 and 70 years (mean age, 57 years). All patients underwent primary tumor resection, and the majority also received lymph node dissection. Patients with a diagnosis of relapse who had received preoperative radiation, chemotherapy or biotherapy were excluded to avoid any alterations in tumor marker determination resulting from treatment. Patients diagnosed with multiple primary cancers in other organs or tissues were also excluded. The study was approved by the ethics committee of The Third Affiliated Hospital of Kunming Medical University, and all patients provided written informed consent and authorization for use of biological specimens. Demographic and clinical data were obtained from the patients' medical records.

A routine histological examination was performed with hematoxylin-eosin staining at room temperature; the stained slices were reviewed independently by three pathologists under a light microscope. Benign lung tissues were collected from a normal section of the lung in patients with a benign pulmonary disease identified by pathologists. All carcinomas were classified in accordance with the 7th edition of the American Joint Committee on Cancer staging system (26).

Samples were processed for immunohistochemical analysis, in order to detect TLR4 and MyD88 expression levels and distribution patterns. Briefly, 4-µm sections of paraffin-embedded tissues were mounted on charged glass slides and baked at 70°C for 1 h. The slides were allowed to cool to room temperature, deparaffinized in xylene and rehydrated in a graded alcohol series. Sections were then microwave-treated for 5 min in citrate buffer (pH 6.0) for antigen retrieval, and endogenous peroxidase activity was blocked by incubation in 0.3% hydrogen peroxide for 20 min at room temperature. Mouse monoclonal TLR4 (cat. no. SAB1404475; Sigma-Aldrich; Merck KGaA) and rabbit monoclonal Myd88 (cat. no. ab133739; Abcam) antibodies were used to detect TLR4 and Myd88 protein expression, respectively, at 1:600 and 1:250 dilutions in PBS; sections were incubated with these antibodies at 4°C overnight. After two washes in PBS, slides were incubated with undiluted rabbit secondary antibodies from a Dako REAL EnVision detection system/Horseradish Peroxidase for rabbit/mouse secondary antibodies kit (cat. no. K5007; Agilent Technologies, Inc.) for 30 min at room temperature, The peroxidase reaction was developed using 3,3′-diaminobenzidine (DAB) chromogen solution (Dako; Agilent Technologies, Inc.). Sections were visualized with DAB and counterstained with hematoxylin for 2 min at room temperature. Sections were scored semi-quantitatively under a light microscope for the extent of the immunoreaction as follows: 0, 0% immunoreactive cells; 1, <20% immunoreactive cells; 2, 20–50% immunoreactive cells; and 3, >50% immunoreactive cells. In addition, the intensity of staining was scored semi-quantitatively as 0, negative; 1, weak; 2, intermediate; and 3, strong. The final immunoreaction score was defined as the sum of both parameters (extent and intensity), and samples were grouped according to the summed score as negative (0), weak staining (1–3), moderate staining (4–6) and strong staining (7–9). Final immunoreaction scores >0 were defined as positive. All slides were evaluated independently for protein expression by three separate observers; slides with an incongruent grading were scrutinized a second time, and a consensus was reached.

The NSCLC cell lines NCI-A549 (human lung adenocarcinoma) and NCI-H226 (human lung squamous cancer) were used in the present study. The H226 cell line was purchased from the Cell Bank of Chinese Academy of Sciences, and the A549 cell line was obtained from the American Type Culture Collection. A549 and H226 cells were respectively grown in DMEM F12 (Gibco; Thermo Fisher Scientific, Inc.) and RPMI-1640 (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc.) and penicillin (1,000 U/ml)/streptomycin (10 mg/ml) (cat. no. SV30010; Hyclone; GE Healthcare) at 37°C in an incubator containing 5% CO₂.

For the dose response analysis, A549 and H226 cells were treated with different concentrations (A549, 5, 10, 15 µM; H226, 5, 10, 20 µM) of curcumin for 72 h in a 37°C incubator containing 5% CO₂. For the analysis of time-dependent response, A549 and H226 cells were treated with curcumin (10 µM) for various durations (0, 24, 48 and 72 h). To investigate the signaling pathway, A549 and H226 cells were pretreated with either DMSO or curcumin (10 µM) for 72 h, followed by a 12-h serum-starvation period. After the aforementioned treatments, cells were sequentially grown in an incubator with 5% CO₂ at 37°C. EGF (100 ng/ml) was added to the cell culture dish at different time points (1.5, 3 and 6 h) after serum starvation. Following the application of the aforementioned treatments, cells were placed in an incubator with 5% CO₂ at 37°C. Then, cells in the logarithmic growth phase were collected to perform western blot analysis.

Cells (A549 and H226) were harvested after drug treatment and lysed in RIPA buffer (cat. no. 89900; Pierce; Thermo Fisher Scientific, Inc.). The lysates were centrifuged at 12,000 × g for 20 min at 4°C, and the supernatants were collected. Protein concentration was measured using the bicinchoninic acid assay (cat. no. 23227; Pierce; Thermo Fisher Scientific, Inc.). For each sample, 50 µg protein lysate was loaded per well, and protein samples were separated by 10% SDS-PAGE and transferred onto polyvinylidene fluoride membranes (cat. no. ISEQ00010; EMD Millipore) by electroblotting. Membranes were pretreated with 8% non-fat dry milk in TBS-Tween-20 (0.05–0.1%) (TBS-T) for 2 h at room temperature, followed by incubation with primary antibodies for 16 h at 4°C. The primary antibodies used targeting the following proteins were diluted at 1:1,000 and were purchased from Abcam: Cyclin B1 (cat. no. ab72), cyclin A1 (cat. no. ab53699), cyclin A2 (cat. no. ab38), cyclin D1 (cat. no. ab6152), cyclin E1 (cat. no. ab3927), cyclin-dependent kinase (CDK)4 (cat. no. ab108357), CDK1 (cat. no. ab18), CDK6 (cat. no. ab124821), p21 (cat. no. ab109520), p15 (cat. no. ab94688), c-Jun (cat. no. ab32137), phosphorylated (p)-c-Jun Ser63 (cat. no. ab32385), JunB (cat. no. ab31421), p-JunB Ser259 (cat. no. ab30628), c-fos (cat. no. 190289), and p-c-fosSer32 (cat. no. ab79319). The epithelial-mesenchymal transition (EMT) antibody sampler kit (cat. no. 9782) was purchased from Cell Signaling Technology, Inc., and the primary antibodies were diluted at 1:1,000. The other primary antibodies, including TLR4 monoclonal antibody (cat. no. 66350-1-Ig), MyD88 monoclonal antibody (cat. no. 66660-1-Ig), the LY96/MD2 polyclonal antibody (cat. no. 11784-1-AP), EGFR rabbit monoclonal antibody (cat. no. 4405) and p-EGFR (Tyr1068) mouse monoclonal antibody (cat. no. 2236), were obtained from ProteinTech Group, Inc. and were diluted at 1:1,000. Following primary antibody incubation, the membranes were washed three times with TBS-T. Subsequently, the membranes were incubated with horseradish peroxidase-labeled secondary antibody (1:10,000; cat. nos. 7076 and 7074; Cell Signaling Technology, Inc.) for 1 h, prior to detection using an ECL reagent (cat. no. RPN2135; GE Healthcare); all blots were exposed for the same duration (5 min). α-tubulin was used as an internal loading control (1:1,000; cat. no. 2125; Cell Signaling Technology, Inc.). Blots were semi-quantified using Gel-Pro Analyzer, version 4.0 (Media Cybernetics, Inc.).

Cell proliferation was determined using the MTS assay. Briefly, 2×10³ cells/well were plated in 96-well culture plates in 150 µl medium; six parallel wells were assigned to each group, as well as a negative control (without cells). Over a 5-day period, 30 µl MTS substrate was added to each well every 24 h and incubated for 2 h at 37°C in the dark. Absorbance was measured at 490 nm for each sample using a plate reader (BMG Labtech). The half maximal inhibitory concentration (IC₅₀) values were calculated from survival curves using the Bliss method (27). All experiments were performed three times independently.

Following treatment with curcumin, cells were harvested using 0.1% trypsin/EDTA and washed three times with cold PBS by centrifugation at 300 × g for 5 min. For cell cycle analysis, cells were fixed in pre-chilled 70% ethanol overnight at 4°C. The fixed cells were collected, washed twice with PBS, suspended in staining buffer (cat. no. C1052-1; Cell Cycle and Apoptosis Analysis Kit; Beyotime Institute of Biotechnology) containing 10 µg/ml propidium iodide (20X, cat. no. C1052-2; Beyotime Institute of Biotechnology) and 100 µg/ml RNase A (50X, cat. no. C1052-2; Beyotime Institute of Biotechnology), and then incubated at 37°C for ≥30 min in the dark to eliminate intracellular RNA. Analysis was performed on a FACSAria system (BD Biosciences) and data were analyzed using CellQuest software (version 5.1; BD Biosciences). All of the samples were assessed three times.

A549 cells were seeded into 6-well plates at a density of 100 cells/well. The medium containing curcumin (5, 10 and 15 µM) and 10% FBS was changed every 3 days. After 2 weeks, colonies resulting from the surviving cells were fixed with 3.7% methanol for 30 min, stained with 10% Giemsa for 15 min at room temperature and counted. H226 cells were plated into 6-well plates at a density of 300 cells/well and were cultured in 10% FBS and medium containing curcumin (5, 10 and 20 µM). Similarly, the medium was also changed every 3 days. After 3 weeks, the developed colonies were fixed with 3.7% methanol and stained with Giemsa. Finally, colonies containing ≥50 cells were counted. Each assay was performed in triplicate.

The Transwell migration assay was used to determine the migratory ability of cells. Briefly, A549 cells were grown in DMEM F12 containing 10% FBS and curcumin (5 and 10 µM). After 48 h, the cells were harvested by trypsinization and washed once with PBS. To measure cell migration, 8-µm pore size culture inserts (Costar; Corning Incorporated) were placed into the wells of 24-well culture plates, separating the upper and the lower chambers. In the lower chamber, 500 µl DMEM F12 containing 10% FBS was added, which was used as a chemoattractant. Subsequently, serum-free medium containing 1×10⁵ cells/well was added to the upper chamber for migration assays. After 7–8 h of incubation at 37°C in an atmosphere containing 5% CO₂, the cells on the upper surface of the membrane were removed using a cotton swab, and the cells that had migrated through the pores were fixed with 3.7% methanol for 30 min and stained with hematoxylin for 5 min at room temperature. Finally, cell migration was determined by calculating the number of migrated cells in three visual fields per well by light microscopy (Olympus Corporation) at ×100 magnification. Each experiment was performed at least three times.

Data are presented as the mean ± standard deviation. SPSS software for Windows (version 18.0; SPSS, Inc.) was used for statistical analysis. The association between TLR4 or MyD88 expression and clinicopathological parameters was analyzed by χ² test, and contingency coefficient test was used to assess the association between TLR4 and MyD88. One-way ANOVA was used to analyze multiple groups, followed by post hoc analysis using Student-Newman-Keuls test for comparison between any two groups or Dunnett's test for comparison among experimental groups and the control group. Log-rank test was used for survival analysis. The hazard ratio was determined using Cox regression. The Kaplan-Meier Plotter (www.kmplot.com) was used to estimate the survival probability of TLR4 and MyD88 in NSCLC. P<0.05 was considered to indicate a statistically significant difference.

Curcumin is a low molecular weight (368.39 Da) polyphenol, the chemical structure of which is shown in Fig. 1A. To investigate the effect of curcumin on NSCLC cell proliferation, the IC₅₀ was determined in A549 and H226 cells from a dose-response curve after 72 h of treatment. The IC₅₀ values of curcumin were 19.71 and 30.88 µM in A549 and H226 cells, respectively (Fig. 1B and C). According to these results, 15 µM curcumin caused a 20–30% decrease in A549 cell survival rate; a similar decrease was found in H226 cells in response to 20 µM curcumin. Subsequently, the MTS assay and colony formation assay were conducted to evaluate the proliferation of NSCLC cells treated with 5, 10 and 15 µM, or 5, 10 and 20 µM curcumin. As shown in Fig. 1D and E, treatment with curcumin for 4 days significantly reduced cell proliferation in a dose-dependent manner. Subsequently, the colony formation assay detected fewer colonies in the curcumin-treated groups compared with in the DMSO group (Fig. 1F and G). These data suggested that curcumin may inhibit cell proliferation in NSCLC.

To better understand the mechanism by which curcumin inhibited cell proliferation, the present study further analyzed the effects of curcumin on cell cycle distribution in A549 and H226 cells by flow cytometry. The proportion of cancer cells in G₂/M phase was significantly increased in the curcumin-treated groups compared with in the DMSO group in A549 cells, and the proportion of cells in G₀/G₁ phase was concomitantly decreased (Fig. 2A and C). However, in H226 cells, 5 and 10 µM curcumin produced very little change in all phases of the cell cycle. However, 20 µM curcumin markedly increased the population of cells in G₂/M phase (Fig. 2B and D). These data indicated that curcumin may modulate G₂/M transition in the cell cycle progression of NSCLC cells.

The cell cycle process is well known to be controlled by several types of checkpoints, including cyclins, CDKs and CDK inhibitors (CDKIs) (28). To explain the molecular mechanism by which curcumin induced G₂/M phase arrest, the expression of various cell cycle checkpoints were detected. The results demonstrated that the protein expression levels of the cell cycle promoters cyclin A1, cyclin A2, cyclin B1, cyclin E1, cyclin D1 and CDK1 were markedly decreased, whereas the expression levels of CDKIs, including p15 and p21, were significantly upregulated (Figs. 2E, and S1A and B). Activator protein-1 (AP-1) has been reported to be closely associated with the proliferation of tumor cells (29), and activation of c-Jun and JunD stimulate cyclin D1 (30,31). Therefore, this study examined the alteration in AP-1 and observed an obvious decrease in the protein expression levels of c-Jun and c-fos, as well as phosphorylation on c-Jun Ser63 and c-fos Ser32 (Figs. 2F and S1C). These findings indicated that curcumin may slow cell cycle progression by downregulating interrelated cyclins and CDKs modulated by c-Jun and c-Fos. However, no significant differences were observed in CDK4, CDK6, JunB and p-JunB Ser259 expression, thus suggesting that these signaling molecules were not affected by curcumin.

Briefly, these data indicated that curcumin may serve a key role in modulating the cell cycle transition in NSCLC via altering the protein expression levels of c-Jun and c-fos, thereby further affecting cell cycle regulators, including cyclins (cyclin A1, cyclin A2, cyclin B1, cyclin E1 and cyclin D1) and CDK1.

A Transwell migration assay was conducted to study the effects of curcumin on NSCLC cell migration. As shown in Fig. 3A and B, the number of migrated A549 and H226 cells in the curcumin-treated group was markedly decreased compared with in the DMSO-treated group, which indicated that curcumin may exert anti-metastatic effects. Subsequently, alterations in EMT-related proteins were detected and E-cadherin was revealed to be markedly upregulated, whereas vimentin was markedly downregulated in the curcumin-treated group (Figs. 3 and S2). Furthermore, the expression levels of transcription factors associated with E-cadherin, including transcription factor 8 (TCF8), Snail and Slug, were determined. In line with our expectations, the expression levels of these transcription factors were markedly decreased in response to curcumin treatment (Figs. 3 and S2). These findings suggested that curcumin inhibited NSCLC cell migration via EMT-related proteins.

TLR4 and EGFR are not only membrane receptors but also upstream effectors of the AP-1 protein (32). The present study detected MD-2, TLR4, MyD88 and EGFR protein expression using western blot analysis. It has previously been reported that curcumin regulates the TLR4/MyD88 signaling pathway by affecting MD-2 expression (33). Consistently, when cultured with curcumin, A549 and H226 cells exhibited a dose-dependent inhibition of MD-2, MyD88 and EGFR protein levels. Notably, TLR4 was significantly inhibited by curcumin in H226 cells; however, only a non-significant downward trend in TLR4 was detected in response to curcumin in A549 cells (Figs. 4A and B, and S3A and B). In addition, a time-dependent significant decrease in TLR4, MyD88 and EGFR protein expression was detected following curcumin treatment in vitro (Figs. 4C and D, and S3C and D). These results indicated that curcumin inhibited the TLR4/MyD88 and EGFR signaling pathways in NSCLC cells.

Studies regarding the interactions between TLR4 and EGFR are emerging; however, the underlying mechanism remains unclear. A lack of a physical association between TLR4 and EGFR has been demonstrated by coimmunoprecipitation assays in mouse macrophage cells, and EGFR requires the participation of Lyn kinase to activate the TLR4 signaling pathway in mouse septic shock (34). Therefore, whether there is a synergistic relationship between EGFR and TLR4 in curcumin-treated NSCLC cells, and whether TLR4/MyD88 is downstream of EGFR, is currently unknown. This study aimed to determine whether cross talk existed between TLR4 and EGFR in curcumin-treated NSCLC cells. For this purpose, A549 and H226 cells were treated successively with DMSO or curcumin (10 µM) and EGF (100 ng/ml). Western blotting revealed that EGF significantly promoted MyD88 expression, and EGF also recovered MyD88 expression in curcumin-treated A549 cells. Conversely, the inhibitory effect of curcumin on TLR4 was slight and the difference was not statistically significant in A549 cells. In addition, no significant difference in TLR4/MyD88 expression was detected in H226 cells. Furthermore, EGF significantly promoted p-EGFR expression and could recover p-EGFR levels in curcumin-treated NSCLC cells (Figs. 4E and F, and S3E and F). These results indicated that the TLR4/MyD88 and EGFR pathways may synergistically regulate cell proliferation and migration in curcumin-treated NSCLC cells.

The present study examined the expression levels of TLR4/MyD88 in 52 NSCLC tissue samples and 49 benign lung tissue samples. The panels presented in Fig. 5 are representative images of TLR4 and MyD88 expression in lung adenocarcinoma and benign lung tissues. TLR4-positive staining and MyD88-positive staining were confined mainly to the cytomembrane and cytoplasm in NSCLC tissue (Fig. 5A and B), whereas benign lung tissue was negatively or weakly stained (Fig. 5C and D). TLR4-positive staining was detected in 38 out of 52 (73.1%) NSCLC samples but only in 10 out of 49 (20.4%) benign lung tissue samples (P<0.001; Table I). Similarly, MyD88-positive staining was detected in 28 out of 52 (53.8%) NSCLC samples but was only found in 12 out of 49 (24.5%) benign lung tissue samples (P=0.003; Table II). However, there was no association between TLR4/MyD88 expression and numerous clinical characteristics, including sex, age, histological type, differentiation, tumor size, lymph node metastasis and pathological stage. Notably, the contingency coefficient analysis between TLR4 and MyD88 expression exhibited a significant positive association (φ=0.308, P=0.026; Table III).

The follow-up period for all 52 patients (I–IIB stage) enrolled in this research began at the discharge date until September 2015. At the end of the follow-up period, six patients were lost to follow-up and 24 had died, leaving 22 survivors. By differentiating patients into positive or negative TLR4/MyD88 staining groups, the prognostic effect of TLR4/MyD88 on the overall survival of patients with NSCLC was analyzed. Kaplan-Meier curve analysis demonstrated that there was no association between TLR4/MyD88 staining status and overall survival (OS) of patients with NSCLC (n=52; P=0.795 and 0.630, respectively; Fig. 6B and D). However, this result differed from the results of Kaplan-Meier data analysis on a larger sample size. As shown in Fig. 6A and C, these results suggested that TLR4/MyD88 expression levels were significantly associated with OS in patients with NSCLC. Of the possible reasons for this disagreement, one may be the small sample size (n=52). Similarly, the Kaplan-Meier Plotter database was used to analyze the association between EGFR expression and OS. The results shown in Fig. S4 suggested that high EGFR expression was closely associated with poor prognosis.