We used a previously described method to obtain LK-A from the leaves of Isodon ternifolius (D. Don) Kudô (15). Briefly, the dried and milled plant material (10 kg) was extracted four times by incubation with 100 l of 70% aqueous Me₂CO for 3 days at room temperature and then filtered. The filtrate was evaporated under reduced pressure, and partitioned with ethyl acetate (EtOAc) (4×60 l). The EtOAc partition (938.5 g) was applied to a silica gel (200–300 mesh), and six fractions, termed A-F, were eluted with CHCl₃-Me₂CO (1:0–0:1). Fraction B (618.5 g) was decolorized on an MCI^® GEL and eluted with 90% methanol-H₂O to yield fractions B1-B4. Fractions B1 (116 g) and B2 (135 g) were further separated by repeated silica gel column chromatography to isolate LK-A (20 g). The LK-A powder was dissolved in dimethyl sulfoxide (DMSO) at a concentration of 50 mM and then stored at −20°C. Before each experiment, we would freshly dilute the LK-A in medium to achieve the working concentrations in this study. The DMSO concentration was kept below 0.1% when used in cell culture and did not exert any detectable effect on cell growth or death. Cell culture reagents, including RPMI 1640 medium, Dulbecco's modified Eagle's medium (DMEM)/F12, recombinant human basic fibroblast growth factor (bFGF), recombinant human epidermal growth factor (EGF) and B-27^® Supplement were purchased from Invitrogen (Thermo Fisher Scientific, Inc., Waltham, MA, USA). The following monoclonal antibodies were used for western blotting: Anti-c-myc (1:1,000; catalog no., 5605S; Cell Signaling Technology, Inc., Danvers, MA, USA), anti-fibronectin (1:2,000; catalog no., 610077; BD Biosciences, Franklin Lakes, NJ, USA), anti-β-actin (1:1,000; catalog no., 66009-1-1g; Proteintech Group, Chicago, IL, USA) and anti-α-tubulin (1:3,000; catalog no., sc-8035; Santa Cruz Biotechnology, Inc., Dallas, TX, USA). All other chemicals, including bovine serum albumin, protease inhibitor cocktail, PBS and Tween-20, were purchased from Sigma-Aldrich (Merck Millipore, Darmstadt, Germany).

S18 and S26 cells are clones of the human NPC cell line CNE2. A stable radioresistant NPC cell line (Sune2-IR) and its parental cell line (Sune2), and the 5–8F NPC cell line, were supplied by and maintained in the State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Sun Yat-sen University Cancer Center (Guangzhou, China) in RPMI 1640 medium supplemented with 5% fetal bovine serum (FBS) (Invitrogen; Thermo Fisher Scientific, Inc.), penicillin (100 U/ml) and streptomycin (100 U/ml). The NPC cell lines were incubated at 37°C in 5% CO₂/95% air.

S18 and S26 cells were plated in triplicate at a density of 200 cells per well in ultra-low attachment 6-well plates (Corning Incorporated, Corning, NY, USA), and then cultured in DMEM/F12 with 20 ng/ml recombinant human bFGF, 20 ng/ml recombinant human EGF and B-27^® Supplement. The spheres were collected after 7 days and counted under a light microscope. Then, we dissociated the primary sphere cells into single-cell suspensions, which were cultured to allow the regeneration of spheres.

SP cell analysis and isolation were performed by fluorescence-activated cell sorting (FACS) (Beckman Coulter, Inc., Brea, CA, USA). Before SP cell analysis, cells were pretreated with different concentrations of LK-A for 48 h. Subsequently, the cells were resuspended at a density of 1×10⁶ cells/ml in RPMI 1640 supplemented with 2% FBS. Then, cells were incubated with 5 µg/ml Hoechst 33342 (Sigma-Aldrich; Merck Millipore) either alone or with 100 µg/ml verapamil (Sigma-Aldrich; Merck Millipore) at 37°C in the dark for 90 min. Cells were washed, centrifuged at 1811 × g for 10 min at 4°C and resuspended in cold PBS. All cells were kept at 4°C in the dark before FACS analysis using dual wavelength analysis.

First, 2,000 cells were seeded into 96-well plates, incubated overnight and then treated with various concentrations of LK-A for 48 h. Then, 20 µl of MTT (5 mg/ml) was added to each well, and the plate was incubated at 37°C for 4 h. Subsequently, the supernatant was carefully removed, and 150 µl/well DMSO was added to dissolve the formazan crystals. The absorbance of the soluble product was measured with a microplate spectrophotometer at 490 nm (µQuant™; Biotek Instruments, Inc., Winooski, VT, USA). This experiment was performed in six replicates and repeated three times. We calculated the percentage of cell viability for each concentration of LK-A by using the following formula: Cell viability (%) = A570 nm (sample) / A570 (control DMSO) × 100. The half maximal inhibitory concentration (IC₅₀) was determined with GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA).

S18 and S26 cells were counted, plated in triplicate at 2,000 cells per well (200 ml) in 96-well plates, and allowed to grow overnight. For the experiment group, a concentration gradient of cisplatin (DDP; Jiangsu Hansoh Pharmaceutical Group Co., Ltd., Lianyungang, China) was added to the wells. For the combination groups, 1.0 µM LK-A was mixed with a concentration gradient of DDP and then added to the wells. Cell viability was measured 48 h later by adding an MTT solution. The observation value was detected at 490 nm. The results from the assays were analyzed for the combination effect between LK-A and DDP according to the method described by Jin (17). The following formula was used: Q = Ea + b / [Ea + Eb - (Ea × Eb)], where Ea + b, Ea and Eb are the average effects of the combination treatment, LK-A only and DDP only, respectively (17). Q<0.85 indicates antagonism, 0.85≤Q<1.15 indicates additivity and Q≥1.15 indicates synergism.

The NPC cells were seeded into six-well plates with 400 cells per well. Twenty-four hours after plating, LK-A was added at a concentration of 0.4 µM. After 24 -h exposure to LK-A, the Sune2 cells and the radioresistant Sune2-IR cells were irradiated at doses of 0, 1, 2 and 3 Gy. Immediately following irradiation, the cells were incubated for 7 days at 37°C to allow colony formation in a 5% CO₂ environment. After 10 days, the colonies were washed with PBS, fixed with pure ethanol and stained with 0.5% crystal violet. Colonies containing >50 cells were counted as clonogenic survivors. Each point on the survival curves represents the mean surviving fraction (SF) from at least three dishes. The equation SF=1-(1-e-^D⁄D0)^N was applied to calculate several parameters, including the cellular radiosensitivity (mean lethal dose, D0), the capacity for sublethal damage repair (quasi-threshold dose, Dq) and the extrapolation number (N). Those parameters were used to calculate the sensitization enhancement ratio (SER) and the SF2 (the SF after irradiation at a dose of 2 Gy).

Cells were seeded into 6-well plates and incubated overnight. Then, the cells were treated with various concentrations of LK-A for 48 h. We used a previously described western blotting method (18). Briefly, equal amounts of protein were separated by 9% SDS-PAGE and electrophoretically transferred onto polyvinyl difluoridine membranes. Rabbit anti-human c-myc antibody (1:1,000; Cell Signaling Technology, Inc.) and mouse anti-human fibronectin antibody (1:2,000; BD Biosciences) were used to detect the expression of c-myc and fibronectin, and were incubated with the membranes for 12 h at 4°C. Anti-rabbit immunoglobulin (Ig)G and anti-mouse IgG secondary antibodies were also used, and α-tubulin was used as an internal control. A chemiluminescence substrate and photographic X-ray imaging was used for visualization.

Data are presented as the mean ± standard deviation. GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA) was used to determine IC₅₀, and SigmaPlot v10.0 (Systat Software, Inc., San Jose, CA, USA) was used to construct cell survival curves. Student's t-test was used to compare the means of various groups. P<0.05 was considered to indicate a statistically significant difference.

To investigate the effect of LK-A on the self-renewal ability of NPC CSCs, we evaluated the sphere-forming capacity after adding LK-A to S18 and S26 cells growing in serum-free non-adherent medium. The result revealed that interference by LK-A resulted in markedly diminished sphere size and number of spheres in a dose-dependent manner in both S18 and S26 cells (Fig. 1A and B).

To further investigate the effect of LK-A on second-generation spheres, after spheres formation, we dissociated the sphere cells into single cells, and we cultivated these single cells in serum-free non-adherent culture with various concentrations of LK-A again. The first-generation spheres were capable of generating second-generation spheres, suggesting that NPC sphere-generated cells have the capacity of self-renewal (Fig. 1C). Meanwhile, we observed that second-generation spheres formed more spheres than primary spheres (Fig. 1B and D). These second-generation spheres were observed to have decreased sphere-forming efficiency under treatment with increasing concentrations of LK-A, with smaller sphere size and lower number of spheres (Fig. 1C and D). These data indicated that LK-A inhibited the self-renewal capacity of NPC cells in a dose-dependent manner.

SP cells are one subpopulation of the cancer cells that effluxes the DNA binding dye Hoechst 33342 out of the cell membrane. Previous studies have shown that the SP cells have stem cell characteristics and enrich the stem cell population in NPC (19). To investigate whether LK-A could impact the percentage of the SP cells in NPC cells, we measured the SP cell percentage of S18 and S26 cells treated with LK-A for 48 h. S18 and S26 are two cell lines originated from CNE2, and S18 has a higher stemness characteristic with a high SP cell proportion, while S26 has a lower stemness characteristic with a lower SP cell proportion (13). As shown in Fig. 2A and B, both in S18 and S26 cells, SP cells are blocked by verapamil at a final concentration of 100 µg/ml. These results indicated that the percentage of SP cells was ~26-fold higher in S18 cells than in S26 cells (38.23 vs. 1.47%). Upon treatment with LK-A, the percentages of SP cells in S18 and S26 cells were all decreased (Fig. 2A and B). S18 cells showed a 2.93-fold decrease in the SP cell proportion upon addition of 0.5 µM LK-A (Fig. 2A). Taken together, these results indicate that LK-A could decrease the percentage of the SP cells in NPC cells.

CSCs are believed to be the cause of chemotherapy resistance (6). Chemotherapy is not effective for a large number of patients with NPC. As LK-A could decrease the proportion of SP cells in NPC cells, we attempted to determine whether LK-A could increase the chemotherapy sensitivity to DDP of NPC cells. As shown in Fig. 3A-C, the dose-response curves of DDP with 1 µM LK-A were obviously shifted to the left in NPC cells. The IC₅₀ value for DDP in the presence of LK-A in S26, S18 and 5–8F cells was significantly reduced (Table I). According to the method by Jin (17), the Q value was 1.04 in S18 cells, 0.86 in S26 cells and 1.02 in 5–8F cells at the concentration of 6.0 µM LK-A (Fig. 3D). As shown in Fig. 3E, at a higher concentration of LK-A (8.0 µM), the Q value was 0.97 in S18 cells, 0.91 in S26 cells and 0.87 in 5–8F cells. To the best of our knowledge, Q<0.85 indicates that the combination of two drugs has an effect of antagonism, while a Q value between 0.85 and 1.15 indicates an additivity effect and Q>1.15 means synergism. Therefore, these data suggested that LK-A has an additive effect to that of DDP in the treatment of NPC cells.

Despite improvements in radiation technology, local recurrence or metastasis occurs in a high proportion of NPC patients due to radioresistance (20). According to the experiment results described above, we aimed to investigate whether LK-A could sensitize the radioresistance of NPC cells. The Sune2-IR radioresistant cell line was obtained from Sune2 cells exposed to ionizing radiation (cumulative dose is 28 Gy). Using the clonogenic survival assay, we found a marked reduction in clone formation in Sune2 and Sune2-IR cells after irradiation combined with the intervention of LK-A (Fig. 4A). Each point on the survival curve represents the mean SF from triplicate experiments (Fig. 4B). There was a significant difference in SF between parental and radioresistant cells at 1, 2 and 3 Gy of radiation, indicating that the Sune2-IR cells were more radiation resistant than the Sune2 cells. Additionally, LK-A (0.4 µM) shifted the Sune2 and Sune2-IR cell dose survival curves clearly to the left, which indicated the increased radiosensitivity of Sune2 and Sune2-IR cells caused by LK-A treatment (Fig. 4B). Sune2 and Sune2-IR cells treated with LK-A (0.4 µM) had a SER of 1.16 and 1.14, respectively (Fig. 4C). The survival curve parameters are listed in Table II. All parameters of the LK-A groups were smaller than those of the control groups.

It was reported that silencing fibronectin extra domain A (EDA) enhances radiosensitivity in NPC (21). Therefore, we performed western blotting to detect the protein expression of fibronectin in the Sune2 and Sune2-IR cell lines. Fig. 5A demonstrates that Sune2-IR cells expressed fibronectin at a higher level than Sune2 cells. These results demonstrated that LK-A could sensitize the radioresistance of NPC cells to irradiation.

As it is well known that there are not reliable CSC markers in NPC cells (22), we investigated whether LK-A could influence the expression of stem cell markers in NPC cells. The expression of Nanog, sex determining region Y-box 2 and octamer-binding transcription factor 4 could not be detected in the S18 or S26 cell lines. In addition, the expression of the Bmi-1, ATP-binding cassette (ABC) G2 and β-catenin were not downregulated by the interference of LK-A (data not show). However, we could detect that LK-A decreased the expression of c-myc in NPC cells (Fig. 5B). c-myc is sufficient to impart high self-renewal and tumor-propagating capacities in breast cancer (23), and overexpression of c-myc increases the stemness of NPC cells (24). These data indicated that LK-A may influence the stemness of NPC cells by downregulating the expression of c-myc.

As aforementioned, LK-A may sensitize Sune2 cells to radiation and reversed the radioresistance of Sune2-IR cells through influencing fibronectin protein expression. Fibronectin is known to play important roles in angiogenesis, lymphangiogenesis and metastasis in malignant tumors (25). Therefore, we wondered whether LK-A treatment could downregulate the protein expression of fibronectin in NPC cells. As shown in Fig. 5C, LK-A treatment downregulated the protein expression of fibronectin in a dose-dependent manner. These data indicated that LK-A may impair radioresistance by downregulation of fibronectin.