The EAC cell lines (OE19 and OE33) were derived from human EAC and were purchased from the European Collection of Cell Cultures (ECACC; Sigma, St. Louis, MO). The OE33 cell line was established from an adenocarcinoma of the lower esophagus arising in Barrett's esophagus and exhibited poor differentiation. The OE19 cell line was established from an adenocarcinoma of gastric cardia/esophageal gastric junction and exhibited moderate differentiation.

Cells were cultured in RPMI 1640 medium supplemented with 2 mM glutamine containing 10% fetal bovine serum (FBS) and antibiotics (100 U/mL penicillin G, 100 μg/mL streptomycin, and 0.25 μg/mL amphotericin) in a humidified atmosphere of 5% CO2/95% air at 37°C.

The effect of 5-FU (Sigma) treatment on cell viability was evaluated by MTT. Briefly, EAC cells were seeded in 96-well-plates at a density of 2,500 cells/well in 200 μL of medium. After seeding, cells were incubated overnight. The following day, cells were treated with different concentrations of 5-FU (1, 10, 50, or 100 μg/mL), and then incubated for 72 h. Next, cells were washed and treated with MTT for at least 2 h. Colorimetric analysis was performed at a wavelength of 570 nm using a standard microplate reader. To determine cell viability, percent viability was calculated as [(absorbance of drug-treated) sample/(control absorbance)] × 100. 5-FU was dissolved in DMSO as a stock solution. All the assays were performed in triplicates, in three independent experiments.

Cells were grown in culture in 175-cm² flasks until they reached 70–80% confluence. Then, cells were treated with 5-FU at IC₅₀ concentration (10 μg/mL). After 72 h of treatment, cells were rinsed with PBS, and the surviving cells were subjected to RNA extraction using an RNeasy Fibrous Tissue Kit (Qiagen, Crawley, Surrey, UK) according to the manufacturer's instructions. The total RNA isolated was purified using RNeasy Mini Elute Cleanup (Qiagen) and quantified by spectrophotometry.

Relative gene expression was determined using the GeXP genetic analysis system (Beckman Coulter, Barcelona, Spain), which allows multiplex detection and quantitation of gene sets in a single analysis (Rai et al., 2009). RT reactions (10 μL) contained 50 ng RNA, 200 nM reverse primers, 2.5 μL kanamycin resistant (Kan^r) RNA, 2 μL 5X RT Master Mix buffer, and 0.5 μL reverse transcriptase. The conditions of RT reactions were: 1 min at 48°C, 5 min at 37°C, 60 min at 42°C, and 5 min at 95°C. Reverse transcriptase, RT master mix buffer, and Kan^r RNA were supplied in Genome Lab GeXP Start Kit. Intron spanning primers were designed using the GenomeLab eXpress Designer software and expression analysis was done with the GenomeLab eXpress Profiling software (Beckman Coulter). Multiplex amplified fragments were separated on a GenomeLab XP capillary sequencer and peak areas were normalized using a reference gene to calculate relative gene expression.

The PCR reaction was performed using 4.65 μL of the 10 μL RT reaction and a final concentration of 200 nM forward primer. 25 mM MgCl₂, 5X PCR buffer, and ThermoTaq polymerase were also added at the concentrations indicated in the Genome Lab GeXP Start Kit. The PCR reaction conditions were: 10 min at 95°C; followed by 35 cycles of 30 s at 94°C, 30 s at 55°C, and 1 min at 68°C. PCR products were analyzed on the GeXP after dilution (800x) with sample loading solution. Kanamycin was used as the RNA internal positive control.

We determined the expression of CD24, CD34, CD44, CD133, ESA, ABCG2, BAX, BCL2L1, and LGR5 genes. For relative gene expression calculation, raw data for each gene were normalized to the housekeeping gene TBP and were subsequently calculated relative to an internal standard (Kan^r gene). The primers are summarized in Table 1. All the experiments were performed in triplicate.

Cells were grown on sterile coverslips placed in 24-well-plates. After 72 h, cells were washed and fixed in methanol for 15 min at −20°C. Cells were then incubated with the primary CD24 antibody (diluted 1/100 in PBS/1% BSA; clone 32D12, Dianova, Hamburg, Germany) or ESA antibody (diluted 1/200; Abcam, MA USA) for 90 min at room temperature in a high-humidity chamber. After washing twice with PBS, the Alexa Fluor 488 goat anti-mouse (1/250 dilution) and Alexa Fluor 488 goat anti-rabbit (1/250 dilution) were added, respectively, and cells were incubated at room temperature for 1 h in a high-humidity chamber.

After several washes, coverglasses were mounted on glass slides using Mowiol. Preparations were observed in a fluorescence microscope (Olympus IX81, Tokyo, Japan).

Control and 5-FU-treated (10 μg/mL) EAC cells were subjected to flow cytometry analyses to determine CD24 and ESA levels. Briefly, the cells were harvested, centrifuged, and suspended in 50 μl PBS. Cells were then incubated with 5 μl mouse anti-human CD24-PE antibody (clone ALB9; Immunotech, Marseille, France) or ESA-FITC antibody (clone ESA214; GeneTex, Irvine, CA) for 15 min at 4°C. Mouse anti-human IgG1-PE and mouse anti-human IgG1-FITC were used as isotype controls for CD24 and ESA, respectively. Samples were analyzed with a FACSAria cell sorter (BD Biosciences, San Jose, CA, USA).

Similarly, tumor spheres and cells treated with celecoxib were subjected to flow cytometric analysis to examine CD24 expression. All the experiments were performed at least three times.

Tumor sphere cultures were established from trypsinized cells from adherent cultures of the parental OE33 and OE19 cell lines. For the formation of esophageal spheres, cells were plated in 25 cm² ultra-low attachment flasks (Corning, Tewksbury, MA, USA) and cultured in serum-free Dulbecco's modified eagle medium (DMEM-F12, Sigma) containing: 5 μg/mL insuline (Sigma), 0.4% bovine serum albumin (Sigma), 10 ng/mL human basic fibroblast growth factor (bFGF; Sigma), 20 ng/mL epidermal growth factor (EGF; Sigma), 2% B27 Supplement (Invitrogen Corporation, Grand Island, NY) and antibiotic/antimycotic (Sigma). Cells were supplemented with fresh medium and growth factors twice weekly. At day 14, spheres >40 μm in diameter were collected using a cell strainer (BD), dissociated to single cells, and reseeded to evaluate self-renewal through the formation of a second generation of tumor spheres. In order to determinate putative CSC markers, an aliquot was analyzed by flow cytometry.

For differentiation assays, disaggregated spheres were seeded in RPMI 1640 medium with 10% FCS in adherent conditions.

In another set of experiments, OE33 cells were plated for the formation of esophageal spheres and then treated with celecoxib (10, 20, or 40 μM) for 7 days. Thus, we determined the size of spheres using Image J software at days 2 and 7 by measuring sphere diameter.

Results are expressed as mean ± SEM. Student t-tests and analysis of variance followed by a Dunnet's test were used to calculate the statistical differences, and p < 0.05 was considered statistically significant.

MTT assay was used to establish IC₅₀-value for 5-FU in our cell lines. IC₅₀ doses were obtained from the dose-response curves (GraphPad Prism 4.0, La Jolla, CA, USA) and were established at 10 μg/mL for OE33 cell line and 10.96 μg/mL for OE19 cell line (Figure 1). The fraction that survived 5-FU exposure was characterized for cancer stem phenotype and examined for enrichment in cancer stem markers.

To characterize whether cells that survived 5-FU exposure were related to the cancer stem phenotype, we compared gene expression of genes involved in cell survival (BAX and BCLL2) and drug resistance (ABCG2) among cells surviving 5-FU exposure and control cells. Using the GeXP genetic analysis system, we determined mRNA expression of the BAX pro-apoptotic gene, the BCLL2 antiapoptotic gene, and the ABCG2 gene in the 5-FU-resistant fraction and in untreated cells. BCLL2 expression was significantly increased in 5-FU treated cells (102% in OE19 cells, p = 0.001 and 51% in OE33 cells, p = 0.002) vs. untreated cells. In contrast, BAX gene expression was significantly reduced in treated cells (28% in OE19 cells, p = 0.002 and 19% in OE33 cells, p = 0.02; Figures 2A,B).

Drug resistance of CSCs is mediated in part by the ATP-binding cassette (ABC) drug transporters. Therefore, we examined ABCG2 expression in these cells. Although levels of ABCG2 mRNA were increased in 5-FU-treated OE19 cells (123% vs. control cells, p = 0.001), no difference in expression was observed in OE33 cells (Figure 2C).

Next, we determined mRNA expression levels of a variety of surface markers related to CSCs (CD24, CD34, CD44, CD133, ESA, and LGR5) in the 5-FU-resistant fraction and untreated cells. Expression of all analyzed markers was observed in both cancer cell lines with the exception of LGR5, which was not detected in the OE33 cell line. CD24, ESA, and CD44 were the main CSC markers expressed. Cells surviving treatment with 5-FU exhibited a significant increase of CD24 mRNA expression (87% in OE19 cells, p = 0.04 and 66% in OE33 cells, p = 0.001) and a moderate increase of ESA mRNA expression (40% in OE19 cells, p = 0.004 and 27% in OE33 cells, p = 0.005) compared with untreated cells. No statistically significant differences were observed when the mRNA expression of CD34, CD133, and LGR5 was analyzed. The level of CD44 mRNA was slightly increased (18%, p = 0.027) in 5-FU treated OE33 cells (Figures 2D,E).

Given that identification of CSC is performed using cell surface markers, and based on the above results, we determined by immunofluorescence the specific cellular localization and by flow cytometry the levels of surface protein CD24 expression and ESA in 5-FU treated cells and in untreated cells.

CD24 staining intensity in esophageal cancer cells was very heterogeneous; that is, not all of the tumor cells displayed similar levels of protein expression (Figures 3A,B). The distribution of CD24 protein was predominantly membranous in OE19 cells (Figure 3A), while membranous and cytoplasmic staining were observed in the OE33 cancer cell line (Figure 3B).

Flow cytometry revealed that the intensity of CD24 expression, measured as mean fluorescence units (MFI) was significantly higher in 5-FU-treated cells than in control cells in both cell lines (OE19 cells p = 0.046; OE33 cells p = 0.035; Figures 4A–E).

ESA immunostaining revealed membranous and cytoplasmic staining in both cancer cell lines (Figures 3C,D), with apparently homogenous staining. No differences were observed in the ESA MFI between control and treated cells (Figure 4F).

Sphere culture has been used as a method for enriching stem cells and evaluating the stem cell self-renewal process. Therefore, we investigated whether EAC cells were able to form spheres in vitro when plated in a serum-free media supplemented with growth factors and in anchorage-independent conditions, which allow stem cells to maintain an undifferentiated status (Singh et al., 2003; Cao et al., 2011; Chen et al., 2012). Both esophageal cancer cell lines had the capacity to form multiple spheres, although OE33 cells showed a greater ability to generate spheres than OE19 cells (Figures 5A,B). Spheres also increased in size over time (Figure 5C). After 14 days under these culture conditions, spheres derived from OE33 cells ranged from 80 to 100 μm in diameter and spheres derived from OE19 cells ranged from 50 to 60 μm in diameter.

To assess the effects of celecoxib on CD24 expression, OE33 cells were incubated for 72 h with celecoxib (10, 20, or 40 μM) alone or in combination with 5-FU (10 μg/mL), and CD24 MFI was analized by flow cytometry. When cells were treated with 5-FU alone, CD24 expression was increased with respect to untreated cells. However, when cells were co-treated with 5-FU and 40 μM celecoxib, we found that celecoxib significantly attenuated 5-FU-induced CD24 expression (p = 0.017). Likewise, cells treated with 40 μM celecoxib alone also showed a reduction of CD24 compared with the corresponding untreated controls (p = 0.034). No effect was observed at the other doses of celecoxib tested (Figures 7A,B).

To evaluate the effects of celecoxib on tumor sphere formation, OE33 cells were grown as tumor spheres and incubated with celecoxib (10, 20, or 40 μM) for 7 days. We monitored the size of spheres in the different experimental conditions (control and treatment) by measuring the diameter of the spheres with a well-defined spherical shape (Figure 8A). After 7 days of treatment, celecoxib (40 μM) significantly inhibited the growth of the spheres compared to untreated (control) cells (p = 0.01).

In some cases, we also observed that the treatment of spheres with celecoxib for 7 days resulted in the disintegration of these spheres compared with the corresponding controls (Figure 8B).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.