The human NSCLC 95-D cell line, which is a commercial cell line, was obtained from the Chinese Academy of Science (Shanghai, China). 95-D cells were cultured with Hyclone™ RPMI-1640 medium (GE Healthcare Life Sciences, Logan, UT, USA) supplemented with Gibco™ 10% fetal bovine serum (FBS; Thermo Fisher Scientific, Inc., Waltham, MA, USA). Cells were incubated at 37°C with 5% CO₂. The 95-D LCSCs, after sorting using a BD fluorescence-activated cell sorting FACSAria flow cytometer (BD Biosciences, San Jose, CA, USA) were culture in neuroblast medium with 20 ng/ml basic fibroblast growth factor (bFGF) (Thermo Fisher Scientific, Inc.), 20 ng/ml epidermal growth factor (EGF) (Thermo Fisher Scientific, Inc.) and B27 supplement (Thermo Fisher Scientific, Inc.).

Oncomine (https://www.oncomine.org/resource/login.html) is a bioinformatics online cancer microarray database aimed at facilitating and promoting the discovery of the functions from genome-wide expression. The differential expression of NELL1 between lung adenocarcinoma and normal counterparts was analyzed using the Oncomine database by searching ‘Gene:NELL1’; ‘Analysis Type: Cancer vs. normal analysis’; ‘Cancer Type: Lung cancer’.

The 95-D LCSCs were accumulated by sphere formation. Basically, the 95-D cells were digested into single cells with 0.25% trypsin and washed with phosphate-buffered saline (PBS). The cells were seeded in ultra-low attachment dishes and cultured in neuroblast medium with 20 ng/ml bFGF, 20 ng/ml EGF and B27 for 7 days to form spheres. The spheres were centrifuged and digested into single cells for further studies.

To analyze the CD133 relative expression in the accumulated LCSCs in spheres, cells were harvested and washed with FACS buffer [PBS containing FBS (1%; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany)] and sodium azide (0.05%; Sigma-Aldrich; Merck KGaA). The CD133-PE/Cy7 (anti-human; dilution 1:100; cat. no. 372810; BioLegend, Shanghai, China) antibody diluted in FACS buffer was added directly to this mixture and incubated for 30 min at 4°C in the dark. Meanwhile the IgG1-PE/Cy7 (anti-human; dilution 1:100; cat. no. 401908; BioLegend) antibody was used as a negative control. Cells were washed and resuspended in FACS buffer. The cells were maintained on ice until the analysis using the FACSCalibur flow cytometer (BD Biosciences) and sorting using FACSAria flow cytometer (BD Biosciences). Results were obtained by analyzing data with FlowJo version 7.6.1 software (FlowJo LLC, Ashland, OR, USA). The results represent the mean value of three independent experiments. The experiments were performed independently three times and a representative is shown.

To prepare the NELL1 expression construct, the NELL1 cDNA fragment (2932 bp, gene accession #BC069674.1) was amplified by PCR in vitro. The PCR products were digested with NheI and SgsI restriction enzymes and inserted into corresponding sites of the PDS023_PL_IRES vector. The PDS023_PL_IRES lentivirus empty vector or vectors for NELL1 were co-transfected with lentivirus packing vector pMDLg/pRRE, RSV-rev and pMD2.G into 293T cells to obtain lentiviral supernatant. The viral supernatant was collected after 48 and 72 h. Wild-type early passage 95-D LCSCs and 95-D cells were incubated with virus-containing medium in the presence of 4 mg/ml polybrene (Sigma-Aldrich; Merck KGaA). Stable cell lines were established after 4 days of blasticidine 5 µg/ml selection.

To investigate cell invasion, 20 µl Matrigel (BD Biosciences) and 80 µl RPMI-1640 medium were added to the upper layer of a Transwell chamber (Corning Inc., Corning, NY, USA). A total of 1×10⁴ 95-D vector (95-D EV) cells, 95-D LCSC vector (95-D LCSCs EV) and 95-D LCSC NELL1-overexpressing cells (95-D LCSCs NELL1) were gently added to the Matrigel medium, while 600 µl PRMI-1640 medium supplemented with 20% FBS was added to the lower layer of the chamber. The chambers were placed in an incubator at 37°C. After 48 h, the chambers were fixed with liquid methanol and stained with 0.5% crystal violet at 37°C for 30 min, and the chambers were washed with PBS several times. Cotton swabs were gently used to remove the cells on the upper layers of the chamber. The cells on the lower layers were counted within 5 randomly observed images using a Nikon microscope (Nikon, Tokyo, Japan) and statistically analyzed.

The colony formation assay was performed as previously described (29). Briefly, a 5% (w/v) base agar (Sigma-Aldrich; Merck KGaA) solution was prepared and autoclaved. For the bottom agar layer, 2 ml of the 0.4% agar/RPMI-1640 bottom agar layer was added to each well of the 6-well plates and was cooled to semisolid status. Single cells (2,000) were seeded in the 0.3% top agar layer into each well. Cells were cultured at 37°C for two or three weeks. Colonies were stained with 0.05% crystal violet (Sigma-Aldrich; Merck KGaA). All assays were performed in triplicate.

For western blot analysis, samples of the 95-D EV, 95-D LCSCs EV and 95-D LCSCs NELL1 cells were harvested. An amount of 30 µg of the total protein samples was prepared, denatured at 100°C for 5 min, and inserted into the 10% gel wells for SDS-polyacrylamide gel electrophoresis (PAGE) and transferred onto polyvinylidine fluoride (PVDF) membranes (Millipore, Darmstadt, Germany). For immunolabeling, the membranes were blocked with 5% milk in Tris-buffered saline with Tween 20 (TBS-T). The membranes were incubated with primary antibodies overnight at 4°C. These antibodies included anti-CD133 (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 64326; Cell Signaling Technology, Danvers, MA, USA), anti-Oct4 (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 2890; Cell Signaling Technology), anti-Sox2 (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 3579; Cell Signaling Technology), anti-NELL1 (rabbit polyclonal antibody, anti-human; dilution 1:500; cat. no. ab197315; Abcam, Cambridge, UK), anti-ABCG2 (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 42078; Cell Signaling Technology), anti-ABCB1 (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 13342; Cell Signaling Technology), anti-ABCC1 (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 14685; Cell Signaling Technology), anti-β-catenin (rabbit polyclonal antibody, anti-human; dilution 1:5,000; cat. no. ab32572; Abcam), anti-shh (mouse monoclonal antibody, anti-human; dilution 1:500; cat. no. sc-365112; Santa Cruz Biotechnology, Inc., Dallas, TX, USA), anti-p-MET (Tyr1349), (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 3133; Cell Signaling Technology), c-MET (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 8041; Cell Signaling Technology), anti-Notch3 (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 5276; Cell Signaling Technology), anti-HES1 (rabbit monoclonal antibody, anti-human; dilution 1:1,000; cat. no. 11988; Cell Signaling Technology) and anti-α-tubulin (mouse monoclonal antibody, anti-human; dilution 1:2,000; cat. no. 3873; Cell Signaling Technology). Then anti-mouse (rabbit polyclonal antibody; dilution 1:5,000; cat. no. ab6728; Abcam) or anti-rabbit (goat polyclonal; dilution 1:5,000; cat. no. ab6721; Abcam) horseradish peroxidase (HRP) conjugated antibodies were used as secondary antibodies and incubated with membranes for 1 h and blots were developed using Electrochemiluminescence (ECL)-Plus Western detection system (Thermo Fisher Scientific, Inc.) for visualization.

Total RNA of 95-D EV, 95-D LCSCs EV and 95-D LCSCs EV cells were extracted using Invitrogen™ TRIzol^® reagent (Thermo Fisher Scientific, Inc.). Samples were treated with TRIzol followed by chloroform and then centrifuged for 10 min at 12,000 × g at 4°C. The supernatant was discarded and the pellet was washed in cold 75% ethanol. Finally, the RNA samples were diluted with 40 µl RNase-free water. A total of 1 µg RNA was reverse transcribed using a Bio-Rad script cDNA synthesis kit (Bio-Rad Laboratories, Hercules, CA, USA). Real-time PCR analysis of CD133, Oct4, Sox2, NELL1, ABCG2, ABCB1 and ABCC1 were performed for quantification using a Mx3000P qPCR system (Stratagene, San Diego, CA, USA) with qPCR cycling conditions (an initial denaturation at 95°C for 1 min and then 45 cycles of 15 sec at 95°C, 31 sec at 60°C). The sequences of primers used for the qRT-PCR are listed in Table I. SYBR^® Premix Ex Taq™ II (Takara Bio, Inc., Tokyo, Japan) was used and 20 µl per gene was analyzed. The relative fold change was quantified by 2^−ΔΔCq (30), and β-actin was used as a housekeeping control.

Cell Counting Kit-8 (CCK-8) was used to conduct the proliferation assays. 95-D EV, 95-D LCSCs EV and 95-D LCSCs NELL1 cells were seeded at 1×10⁴ cells per well in 96-well microtiter plate and were maintained at 37°C for 24 h. A mixture of 200 µl RPMI-1640 medium with different concentrations of carboplatin and cisplatin was added into each well. After 48 h of incubation, a mixture of 190 µl RPMI-1640 with 10% FBS and 10 µl of CCK-8 was added into each well and measured at 450 nm. Each experiment was performed in replicates of six and background reading of the media was subtracted.

All of the experiments were performed three times independently. Statistical analysis (simple or plural) was performed using Statistical Package for Social Science (SPSS) software, version 19 (IBM Corp., Armonk, NY, USA). Groups were compared with Student's t-test or one-way analysis of variance (ANOVA) with Tukey's honestly significant difference (HSD) post hoc test where groups were more than two. Results were considered to be statistically significant at P<0.05.

To evaluate how transcription of NELL1 correlates with NSCLC, we analyzed clinical cohorts of NSCLC patients using the Oncomine database. It was found that NELL1 expression was lower in lung adenocarcinoma (Fig. 1A). Previous research demonstrated that NELL1 plays an important role in osteogenic differentiation (31). Thus, we aimed to ascertain whether overexpression of NELL1 could induce the differentiation of LCSCs. To measure the effect of NELL1 on LCSCs, specific medium was utilized to accumulate the LCSCs in spheres. The 95-D cancer stem-like cells were accumulated by sphere formation. To test whether the formed spheres were indeed cancer stem cells, flow cytometry was used to examine the expression of stemness gene CD133. It was found that CD133 was highly expressed in the sphere cells (Fig. 1B). These highly expressed CD133 cells were next obtained through flow cytometric cell sorting of the CD133-stained cells and these cells were termed 95-D lung cancer stem-like cells (95-D LCSCs). Then, the expression levels of stemness genes (CD133, Oct4 and Sox2) was assessed using western blot analysis. It was found that these genes were highly expressed in LCSCs (Fig. 1C). The results were confirmed using qPCR (Fig. 1D). These results demonstrated that 95-D LCSCs express tumor stem-like cell-related genes and may display tumor stem cell characteristics.

To test whether NELL1 inhibits 95-D LCSC growth, 95-D cells were transfected with a lentivirus empty vector and 95-D LCSCs cells with a lentivirus empty vector or a lentivirus carrying NELL1. The transfection efficiency was confirmed (Fig. 2A and B). NELL1 was highly expressed in the 95-D LCSCs NELL1 group. To test whether the isolated 95-D LCSCs could maintain high expression of CD133 in different passages, the CD133 expression was evaluated in early and multiple passages. We found that CD133 was highly expressed in the different passages (Fig. 2C). Then, we examined the effect of NELL1 on 95-D LCSCs. Transwell invasion assay was used to examine the change in the invasive ability. Overexpression of NELL1 reduced the invasive capability of the 95-D LCSCs compared with the control cells, and the number of invasive cells was significantly decreased (Fig. 2D and E). Overexpression of NELL1 also significantly inhibited the colony formation and growth of 95-D LCSCs (Fig. 2F). These results indicate that NELL1 expression decreased the colony formation and invasion of 95-D LCSCs.

One of the major points of cancer stem cells is their chemoresistance. Carboplatin and cisplatin are common chemotherapeutic drugs. To investigate the role of NELL1 expression in chemotherapeutic drug resistance, a CCK-8 assay was performed (Fig. 3A and B). Our results showed that the chemosensitivity of 95-D LCSCs NELL1 cells was significantly increased compared with the 95-D LCSCs. Subsequently, we analyzed the protein levels of common multi-drug resistance markers ABCG2, ABCB1 and ABCC1. As shown in Fig. 3C and D, the expression of ABCG2, ABCB1 and ABCC1 was significantly decreased in the 95-D LCSCs NELL1 cells. Upon mRNA analysis, the same trends for ABCG2, ABCB1 and ABCC1 protein expression were also identified (Fig. 3E). These data show that overexpression of NELL1 in 95-D LCSCs cells increased the chemosensitivity of these cells.

From the aforementioned results, we confirmed that 95-D LCSCs expressed stem cell genes, CD133, Oct4 and Sox2, and overexpression of NELL1 could inhibit 95-D cell growth. However, the mechanism remained unclear. Studies have shown that Sonic Hedgehog and Wnt pathways are important in cancer stem cells (32,33). We assessed Sonic Hedgehog (Shh) and β-catenin which are important proteins in these two pathways (34,35). Yet, we did not find any change of expression in these two proteins (Fig. 4A). c-MET was found to be activated in cancer stem cells (36,37). We examined the expression of these proteins. As shown in Fig. 4B and C, we found that NELL1 overexpression downregulated p-MET. Previous research has demonstrated that Notch signaling plays an important role in cancer stem cells (38). In the present study, it was also demonstrated that Notch3 and HES1 were downregulated in the 95-D LCSCs NELL1 cells. These results demonstrated that NELL1 may affect 95-D LCSC growth through c-MET/Notch signaling. In addition, overexpression of NELL1 decreased CD133, Oct4 and Sox2 at the protein and mRNA levels (Fig. 4D-F). Based on these data, we conclude that NELL1 may induce differentiation of 95-D LCSCs by inhibiting c-MET/Notch signaling (Fig. 5). Our results showed that NELL1 could be a potential target for lung cancer stem-like cells.