The curative effect of gefitinib is closely related to EGFR mutations such as the deletion mutation of exon-19. The NSCLC cell lines HCC827 and H1650, which contain the deletion mutation of exon-19, were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA), and the established method was in accordance with Engelman et al (28). The cell lines HCC827/G, H1650/G resistant to gefitinib and NSCLC cell lines HCC827 and H1650 were cultured in RPMI-1640 medium (Hyclone, Salt Lake City, UT, USA) with 10% fetal bovine serum (FBS; Gibco, Rockville, MD, USA), penicillin and streptomycin (100 U/ml and 100 mg/ml, respectively; Gibco) in an atmosphere of 5% CO₂ at 37°C.

The cell growth and viability were measured by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay based on the manufacturer's instructions. In brief, the cells were plated in the 96-well plate and the density was 1×10⁴ cells/well. Then, MTT was diluted in phosphate buffered saline to 5 g/l, and 20 µl was added to each well to replace the culture medium. Thereafter, the 96-well plate was stored at 37°C for 4 h. Subsequently, dimethyl sulfoxide (DMSO) was added (150 µl/well) in the plate to dissolve the crystals. Ultimately, absorbance was detected by the microplate reader at 490 nm (Thermo Fisher Scientific, Waltham, MA, USA).

Annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) apoptosis detection kit (Invitrogen, Carlsbad, CA, USA) was used according to the manufacturer's instructions to identify cell apoptosis. Cells were suspended in binding buffer, 10 µl of Annexin V-FITC solution was added, and they were incubated at 4°C for 25 min. Next, cells were washed with PBS, 10 µl of PI was added, and they were incubated for 5 min. Cellular apoptosis was measured by a FACS analyzer (Thermo Fisher Scientific, Waltham, MA, USA).

The full-length cDNA of PAD4 (accession no. NM_012387) and Elk1 (accession no. AB016194.1) were both amplified by reverse transcription PCR, and then cloned into the pCMV-2a/2b vector (Novagen, Madison, WI, USA). The restriction sites included EcoRI and BamHI in the amplified cDNA and pCMV-2a/2b vector. Next, the two different recombinant plasmids were separately transformed into DH5α competent cells (Takara Biotechnology, Dalian, China), and the transformed DH5α was grown overnight at 37°C. Plasmids were extracted and then sequenced to identify the correct ones. The correct plasmids were named as pCMV-2a/2b-PAD4 and pCMV-2a/2b-Elk1.

HCC827/G and H1650/G (5×10⁴/well) were separately plated in the 24-well plate and incubated in a humidified atmosphere containing 5% CO₂ at 37°C for 24 h followed by the transfection strictly according to the manufacturer's instruction. A total of 3 µl TurboFect (Thermo Fisher Scientific) was separately added to the recombinant plasmids (pCMV-2a/2b-PAD4, pCMV-2a/2b-Elk1or pCMV-2a/2b), PAD4 siRNA (5′-GCGAAGACCTGCAGGACAT-3′) or non-specific siRNA with 100 µl serum-free RPMI-1640 medium, and the mixtures were each added to the wells. Finally, the plate was placed in the incubator (Thermo Fisher Scientific) with 5% CO₂ at 37°C for another 24 h.

TRIzol reagent (Takara Biotechnology) was used to extract the total RNA from the cells with the different treatments. A total of 6 µg of the extracted RNA was synthesized into the cDNA that would be used as the template in the qRT-PCR. Additionally, the synthesis process of the cDNA was in terms of the reverse transcription kit (Invitrogen). A volume of 10 µl SsoFast™ EvaGreen Supermix (Bio-Rad, Hercules, CA, USA) was added into 20-µl reactive volume in the qRT-PCR. The qRT-PCR protocol was as follows: 95°C for 20 sec; 35 cycles of denaturation at 95°C for 30 sec, annealing at 52°C (Elk1), 58°C [PAD4, glyceraldehyde-3-phosphate dehydrogenase (GAPDH)] or 60°C (E-cadherin, α-catenin, vimentin and N-cadherin) for 20 sec and extension at 72°C for 30 sec. The data obtained were calculated by 2^−ΔΔCt and the target gene relative expression levels were normalized to GAPDH. The primers are presented in Table I.

The proteins were extracted from the cells using lysis buffer with phenylmethanesulfonyl fluoride (PMSF) for 30 min on ice, and the concentrations of the proteins were measured by the BCA kit (Pierce, Rockford, IL, USA). Twenty-five micrograms of protein was loaded in 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred by a semi-dry blotting apparatus (Bio-Rad). Blots were washed with Tris buffered saline with Tween (TBST), and then blocked using 5% skim milk in Tris buffered saline for 2 h. Next, blots were incubated with the primary antibodies (Cell Signaling Technology, Danvers, MA, USA), diluted (1:500) in the blocking buffer at 4°C overnight, incubated in the horseradish peroxidase-conjugated secondary antibody (Abcam, Cambridge, UK) and diluted (1:1,000) in the blocking buffer at room temperature for 1 h. Blots were analyzed in the Bio-Rad ChemiDoc apparatus.

The differences were determined by Student's t-test for two groups or by one-way ANOVA followed by Bonferroni's test for multiple groups. A P<0.05 was recognized as statistically significant. Data were processed as mean ± standard deviation (SD).

Drug resistance was demonstrated using the Annexin V-FITC/PI apoptosis detection kit and MTT assay. The detections were performed following the cell treatments. The treated cells were cultured in 180 µl RPMI-1640 medium containing 10% FBS and 20 µl of various concentrations of gefitinib for 72 h at 37°C. In the MTT assay, the percent survival (Fig. 1A) was calculated as: (mean absorbance of six replicate wells containing drugs - mean absorbance of six replicate background wells)/(mean absorbance of six replicate drug-free wells - mean absorbance of six replicate background wells) ×100. The results showed that the half maximal inhibitory concentration (IC₅₀) of HCC827/G and H1650/G cell lines were 16.5±6.3 and 14.1±3.7 µmol/l, respectively, and the IC₅₀ of HCC827 and H1650 cell lines were 0.03±0.02 and 0.02±0.01 µmol/l, respectively. The drug resistance of HCC827/G or H1650/G was significantly higher than that of HCC827 or H1650 (P<0.01). In the Annexin V-FITC/PI apoptosis detection assay, the cells cultured with 1 µmol/l gefitinib were measured, and the results indicated that the apoptosis induced by gefitinib in HCC827/G or H1650/G was lower than that in HCC827 or H1650, respectively (Fig. 1B). In conclusion, the establishment of drug resistant cell lines to gefitinib was successful. Additionally, the relative protein expression of PAD4 in HCC827/G or H1650/G was distinctly reduced compared with HCC827 or H1650 (Fig. 2).

To investigate the influence of PAD4 expression changes on the drug resistance of NSCLC cell lines to gefitinib, we promoted and inhibited the expression of PAD4, and cultured the cells with 1 µmol/l of gefitinib. The detection showed that the mRNA (Fig. 3A) and protein (Fig. 3B–D) expression levels of PAD4 in the PAD4 siRNA transfected group were significantly lower than in the non-specific siRNA transfected group. The mRNA (Fig. 3A) and protein (Fig. 3B–D) expression levels of PAD4 in the pCMV-2a/2b-PAD4 transfected group were significantly upregulated contrasting with the pCMV-2a/2b transfected group. Subsequently, the effect of PAD4 overexpression or inhibition on drug resistance was measured by MTT assay and Annexin V-FITC/PI apoptosis detection. In the MTT assay, the cell growth and viability of the pCMV-2a/2b-PAD4 transfected group was obviously decreased in contrast with the pCMV-2a/2b transfected group (Fig. 4A). Conversely, inhibition of PAD4 by transfection of PAD4 siRNA exhibited the opposite effect (Fig. 4A). Moreover, in the Annexin V-FITC/PI apoptosis detection, pCMV-2a/2b-PAD4 transfection induced apoptosis compared with the pCMV-2a/2b transfection (Fig. 4B). Additionally, PAD4 siRNA transfection had the opposite effect of pCMV-2a/2b-PAD4 transfection (Fig. 4B). To summarize, upregulation of PAD4 suppressed drug resistance of NSCLC cell lines to gefitinib.

We identified that Elk1 expression and the EMT activity were both inhibited by PAD4 overexpression in the cells cultured with 1 µmol/l of gefitinib. The mRNA (Fig. 5A and D) and protein (Fig. 5B, C, E and F) of Elk1, N-cadherin and vimentin were significantly downregulated by pCMV-2a/2b-PAD4 transfection while E-cadherin and α-catenin were upregulated. In addition, PAD4 siRNA transfection had a reverse impact compared with pCMV-2a/2b-PAD4 transfection (Fig. 5).

We demonstrated that PAD4 overexpression inhibits both EMT and Elk1. Moreover, it has been reported that Elk1 can regulate the process of EMT (22). So we hypothesized that PAD4 upregulation limits EMT by decreasing Elk1 expression. To this end, we overexpressed both Elk1 and PAD4 by transfecting the plasmids pCMV-2a/2b-Elk1 and pCMV-2a/2b-PAD4 and cultured the cells with 1 µmol/l of gefitinib. The results demonstrated that Elk1 and PAD4 were highly expressed (Fig. 6) and the overexpression of Elk1 apparently reversed the downregulated influence of PAD4 upregulation on EMT (Fig. 6). Furthermore, Elk1 overexpression evidently abolished the drug resistance suppression of PAD4 overexpression, so the cell growth and viability were distinctly increased compared with PAD4 overexpression only (Fig. 7). Above all, PAD4 overexpression inhibited the drug resistance by controlling the expression of Elk1.