The CXCL5/CXCR2 axis contributes to the epithelial-mesenchymal transition of nasopharyngeal carcinoma cells by activating ERK/GSK-3β/snail signalling

Cell culture

NP69 immortalized nasopharyngeal epithelial cells were maintained in keratinocyte/serum-free medium (Invitrogen, Grand Island, NY, USA). Four human NPC cell lines (S18, S26, 5-8F and 6-10B) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, Carlsbad, CA, USA) supplemented with 10% foetal bovine serum (FBS; Gibco, Grand Island, NY, USA). S18 and 5-8F are considered to be highly metastatic cell lines, whereas S26 and 6-10B are considered poorly metastatic cell lines according to previous studies [26, 27].

Patients and tissue specimens

Prior written patient consent and Institutional Research Ethics Committee approval were obtained. No patients received treatment prior to biopsy.

Fifty NPC specimens and fifty chronic nasopharyngitis samples were obtained from patients who underwent nasopharyngeal biopsies before treatment at Sun Yat-sen University Cancer Center (SYSUCC) between November of 2009 and December of 2009.

Enzyme-linked immunosorbent assay (ELISA)

The serum CXCL5 levels of the 100 aforementioned patients (50 patients with NPCs and 50 patients with chronic nasopharyngitis) were measured with a commercially available ELISA kit (Quantikine Human ENA-78; R & D Systems, Minneapolis, MN, USA) in accordance with the manufacturer’s instructions. The secretions of CXCL5 into the media collected from cultures of the NPC or NP69 cells were also determined by ELISA. The ELISA was performed as described previously [28, 29].

Immunohistochemical staining (IHC) and histologic evaluation

Paraffin-embedded blocks were sectioned for IHC staining as performed previously [30]. Briefly, paraffin-embedded tissue specimens were deparaffinized and rehydrated. Antigenic retrieval was performed with citrate buffer (pH 6.0) using a 700-W microwave for 10 min. The sections were then incubated in H₂O₂ (3%) for 10 min, blocked in goat serum at room temperature for 30 min and incubated with anti-CXCL5 (1:100) or anti-CXCR2 (1:200) antibodies overnight at 4 °C. An EnVision kit (DAKO, Carpinteria, CA, USA) was used to detect the primary antibodies followed by 3, 3-diaminobenzidine substrate visualization and counterstaining with haematoxylin. The intensity of IHC staining in the tumour cells was scored independently by two pathologists using the semiquantitative immunoreactive score (IRS) scale (for staining intensity: no staining = 0; weak staining =1; moderate staining = 2; and strong staining = 3. For the percentage of stained cells, 0% = 0; 1%–10% = 1; 11%–50% = 2; 51%–80% = 3; and 81%–100% = 4). The staining score for each tissue was calculated by multiplying the staining value by the percentage category value, and the average of the scores from the two referees was used as the final score. The median and range of the IRS for each antibody and the definitions of high-expression and low-expression of CXCL5 and CXCR2 used in this study are listed in Additional file 1: Table S1.

Plasmids

The psi-LVRU6GP-shRNA-CXCR2, psi-LVRU6GP-shRNA-CXCL5, pEZ-Lv105-CXCR2 and pEZ-Lv105-CXCL5 lentiviral vectors were purchased from GeneCopoeia, Inc., and the target short hairpin RNA (shRNA) sequences are listed in Additional file 2: Table S2. The psi-LVRU6GP and pReceiver-Lv105 lentiviral vectors were used as controls. A small interfering RNA (siRNA) was used to knock down Snail and was constructed and synthesized by Guangzhou RiboBio Co., Ltd. (Guangzhou, China); the target sequences are listed in Additional file 2: Table S2. The pcDNA3.1-GSK-3βS9A plasmids were purchased from Shanghai GeneChem Co., and pcDNA3.1 plasmids were used as a control. All transfections have been described previously [31].

Western blot analysis and immunofluorescence assay

The western blotting and immunofluorescence assays were performed as previously described [32–34]. The primary antibodies and dilutions are listed in Additional file 3: Table S3. The grey values of the target bands were quantified with ImageJ software.

qRT-PCR

qRT-PCR was performed using a LightCycler 480 instrument (Roche Diagnostics) and SYBR Premix Ex Taq (TaKaRa) according to the manufacturer’s instructions. To quantify the cancer metastases in the mouse lungs, specific primers for human HPRT that do not cross-react with its mouse counterpart were designed as described [35–37]. The primers used are listed in Additional file 2: Table S2.

Cell growth and foci formation assays

For cell growth assay, cells were seeded in 96-well plate at a density of 1 × 10³ per well and cell growth rate was assessed by Cell Counting Kit-8 (Dojindo). Cellular growth curves were plotted by using the cellular viability values. For foci formation assay, 1 × 10³ cells per well were seeded dispersedly in 6-well plate. After one-week culture, cell colonies were counted by crystal violet staining. The results are expressed as mean ± SD of three independent experiments.

Migration and invasion assays

Migration and invasion assays were used to evaluate the migration and invasion abilities of cells. Transwell assays were performed using a Boyden chamber containing 24-well transwell plates (BD Inc.) with 8-μm membrane pores. All experiments were performed in duplicate and repeated three times. For the Transwell migration assay, 2 × 10⁴ (S18 and 5-8F) or 4 × 10⁴ (S26 and 6-10B) cells in 400 μl of serum-free DMEM were added to the cell culture inserts without extracellular matrix coating. DMEM containing 10% FBS was added to the bottom chamber. After approximately 18 h of incubation, the cells on the lower surface of the filter were fixed, stained, and examined using a microscope. For the Transwell invasion assay, the membrane was coated with 50 μl of 1:5 diluted Matrigel (BD Biosciences). After the Matrigel had solidified at 37 °C overnight, 4 × 10⁴ (S18 and 5-8F) or 1 × 10⁵ (S26 and 6-10B) cells in 400 μl of serum-free DMEM were added to the cell culture inserts, while the lower chamber was filled with DMEM containing 10% FBS. The Boyden chamber was then incubated at 37 °C with 5% CO₂ for approximately 24 h. The subsequent staining and observation procedures were the same as for the migration assays.

Luciferase reporter assay

The indicated stable cells were seeded in Cignal Finder 10-Pathway Reporter Array plates (QIAGEN, Dusseldorf, GER) when the cell density and cell viability met the transfection conditions. The protocol of the Luciferase Cignal Finder Reporter Array Plate Format Handbook was followed for developing the assays. A Dual-Glo Luciferase Assay System (Promega, Madison, WI) was used to measured dual-luciferase signals after reverse transfection. Three independent experiments were performed, and the calculated means and standard deviations are presented.

Annexin V/Propidium Idodide (PI) double-staining assay

Annexin V/PI staining was performed using the Annexin V-fluorescein isothiocyanate apoptosis detection kit. Cells (3.0 × 10⁵ per mL) were seeded into six-well plate with 2 mL in each well, then treated with 50 nM (the same concentration used in the migration assay) of trametinib for 24 h. Both floating and attached cells were collected, washed with ice-cold phosphate buffered saline (PBS) twice, then incubated at room temperature in the presence of media binding reagent and Annexin V-FITC for 15 min in the dark. After washing with PBS, the cells were resuspended in ice-cold 1 × binding buffer and treated with 10 μL propidium iodide (30 μg/ml) on ice in the dark. Apoptosis was quantified by flow cytometry (Becton Dickinson) at the wavelength of 488 nm immediately and analyzed by the Cell-Quest software.

Animal experiments

All animal experiments were performed in accordance with protocols approved by the Research Animal Resource Center of Sun Yat-sen University. Animals were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). A lung metastasis model was used [35]. Briefly, the indicated stable cells were harvested and washed twice with PBS. Approximately 1 × 10⁶ cells resuspended in 150 μl of PBS were injected into the tail veins of 4-week-old male athymic mice. All mice were sacrificed at 7 weeks after injection. The lungs of the mice were weighed, and the number of metastatic nodules in the lungs was counted. The harvested lungs were fixed in formalin (4 μmol/L). Haematoxylin and eosin (H & E) staining was performed for histologic assessment. Total RNA was extracted for qRT-PCR analysis of hHPRT mRNA expression.

Statistical analysis

Statistical analyses were performed using SPSS 19.0 (IBM, Chicago, IL, USA). Quantitative data between groups were compared using Student’s t-test. Categorical data were analysed by a chi-square test or Fisher’s exact test. Actuarial rates of survival were calculated using the Kaplan-Meier method, and differences were compared using log-rank tests. Multivariate analysis was performed using an adjusted Cox proportional hazards model. A 2-tailed P value less than 0.05 was considered significant.

CXCL5 and CXCR2 are upregulated in NPC tissues and highly metastatic NPC cell lines, and CXCL5 is significantly increased in the sera of NPC patients

The expression of both CXCL5 and CXCR2 was higher in the NPC primary tissues than in the non-tumour tissues at the protein level as quantified by IHC, and the expression of CXCL5 was primarily localized to the tumour cells rather than the mesenchymal portion of the cancerous tissue (Fig. 1a). ELISA analyses also revealed that the serum CXCL5 level was significantly higher in NPC patients than in the non-tumour patients (P < 0.001) (Fig. 1b). Next, we found that CXCL5 and CXCR2 were overexpressed in the NPC cell lines compared with the immortalized nasopharyngeal epithelial cell line NP69. The overexpression of CXCL5 and CXCR2 in the highly metastatic cell lines (S18 and 5-8F) was much more obvious than that in the poorly metastatic cell lines (S26 and 6-10B) (Fig. 1c and d). The levels of CXCL5 in the supernatants of the NP69 and NPC cell lines as detected by ELISA were consistent with the results from the qRT-PCR and western blotting (Fig. 1e).

Clinical significance of the high expression levels of CXCL5 and CXCR2 in NPC

As shown in Table 1, a high expression level of CXCL5 in the NPC primary tissues was significantly associated with nodal classification and progression, and a high expression level of CXCR2 was significantly associated with nodal classification and distant metastasis. As shown in Fig. 1f, a higher level of CXCL5 expression was significantly correlated with poorer overall survival (OS) and distant metastasis-free survival (DMFS), and a higher level of CXCR2 expression was significantly correlated with lower DMFS but not with poorer OS.

Overexpression of the CXCL5/CXCR2 axis promotes NPC cell migration and invasion in vitro and increases lung metastasis in vivo

The S26 and 6-10B cell lines were transfected with CXCL5 and/or CXCR2 expression plasmids to upregulate the expression of CXCL5 and/or CXCR2. Cells transfected with blank vector were used as the control (vec). qRT-PCR and western blotting were conducted to examine the mRNA and protein levels of CXCL5 and/or CXCR2 in the S26 and 6-10B stable cell lines. As shown in Fig. 2a and b, transfection with the CXCL5 and/or CXCR2 expression plasmids caused a significant increase in the mRNA and protein expression of CXCL5 and/or CXCR2 in the S26 and 6-10B stable cell lines when compared with the vec group. ELISA also revealed significantly increased levels of CXCL5 protein in the stable CXCL5 overexpressing cell lines (Additional file 4: Figure S1). The results of the cell growth and foci formation assays revealed that the overexpression of CXCL5 and/or CXCR2 did not influence NPC cell growth, tumour formation or cell proliferation (Fig. 2c and d). However, we found that the cells expressing high levels of CXCL5/CXCR2 together exhibited greater migration and invasion potentials than the other stable cell lines (Fig. 2e).

Next, we developed an in vivo lung metastasis model by injecting S26 or 6-10B cells and their derived cells into the lateral tail veins of nude mice for 7 weeks. As assessed by H & E staining (Fig. 2f and g), overexpression of the CXCL5/CXCR2 axis resulted in significant increases in lung weight and the number of metastatic pulmonary nodules (Fig. 2h and i). Additionally, the individual tumour burden per lung was measured by qRT-PCR using primers specific for the human housekeeping gene HPRT, which detects human but not mouse HPRT mRNA. The expression level of human HPRT in the lungs that had been infiltrated with NPC cells expressing high levels of CXCL5/CXCR2 together were higher than those in the lungs that had been infiltrated with cells expressing high levels of CXCL5 or CXCR2 alone and low levels of CXCL5/CXCR2 together (Fig. 2h and i). Collectively, these results indicated that the overexpression of the CXCL5/CXCR2 axis increases distant metastasis of NPCs in a mouse model.

Silencing the CXCL5/CXCR2 axis inhibits NPC cell migration and invasion in vitro and decreases lung metastasis in vivo

The S18 and 5-8F cell lines were transfected with CXCL5-specific and/or CXCR2-specific shRNAs to decrease the expression of CXCL5 and/or CXCR2. Nonspecific shRNA was used as a negative control (shc). We found that the mRNA and protein levels of CXCL5 and/or CXCR2 were significantly decreased after transfection with CXCL5-specific and/or CXCR2-specific shRNA (Fig. 3a and b). The results were consistent with those obtained using an ELISA to determine the levels of CXCL5 in the cell culture supernatants (Additional file 5: Figure S2). As expected, in both cell lines, CXCL5-silencing and/or CXCR2-silencing significantly attenuated the migration and invasion abilities without influencing cell proliferation (Fig. 3c-e).

We developed an in vivo lung metastasis model by injecting S18 or 5-8F and their derived cells into the lateral tail veins of nude mice for 7 weeks. The results indicated that the silencing of CXCL5 and/or CXCR2 could significantly decrease distant metastasis of NPC cells (Fig. 3f-i).

Overexpression of the CXCL5/CXCR2 axis promotes the metastasis of NPC cells through the induction of the EMT

As shown in Fig. 4a, the NPC cells that expressed high levels of only one of the proteins or neither protein presented with the typical cobblestone-like appearance of normal epithelium, whereas the S18-shc and S26-CXCR2-CXCL5 cells with high CXCL5/CXCR2 expression levels both took on spindle-like, fibroblastic morphologies. The influence of CXCL5/CXCR2 expression levels on cell morphology was further confirmed by immunofluorescent staining for phalloidin (Fig. 4b). Western blotting indicated that the S18-shc and S26-CXCR2-CXCL5 cells exhibited the typical EMT phenotype, including the downregulation of E-cadherin and the upregulation of Vimentin (Fig. 4c and d). Taken together, these data suggested that the overexpression of the CXCL5/CXCR2 axis could promote the EMT process.

Snail is required for CXCL5/CXCR2 axis-triggered EMT and invasion capacity of NPC cells

To further characterize the transcriptional factors involved in the EMT triggered by the CXCL5/CXCR2 axis, we examined the expression profiles of ZEB1, Twist, Slug and Snail in NPC cells. As shown in Fig. 5a, we found significantly increased levels of Snail in the CXCR2- and CXCL5-transfected S26 cells and decreased levels of Snail in the shRNA-CXCR2- and/or shRNA-CXCL5-treated S18 cells. However, we observed no significant changes in the ZEB1, Twist, or Slug levels. Next, after the introduction of siRNA specific for Snail into the S18-shc and S26-CXCR2-CXCL5 cells, we found that E-cadherin expression was enhanced, and Vimentin was downregulated (Fig. 5a). Additionally, the spindle-like, fibroblastic morphology appeared to be abolished, and the migration and invasion abilities of these cells were decreased compared with the control cells (Fig. 5b and c). These results indicated that the CXCL5/CXCR2 axis-induced EMT and invasive capacity of the NPC cells were dependent on the transcriptional factor Snail.

The CXCL5/CXCR2 axis leads to the activation of GSK-3β by phosphorylation

GSK-3β and nuclear factor-κB (NF-κB) have previously been demonstrated to mediate the stabilization of endogenous Snail [20, 41]. Therefore, we evaluated the levels of P65 NF-κB and GSK-3β and their phosphorylation statuses in NPC cells. Western blotting revealed no significant difference in the level of activated P65 between these cells (Additional file 6: Figure S3). However, the phosphorylation of GSK-3β was notably enhanced in the NPC cells with high expression levels of both CXCL5/CXCR2 together, and the expression of Snail changed in parallel with p-GSK-3β (Fig. 6a). These findings indicated that the effect of the CXCL5/CXCR2 axis on the stabilization of Snail in the NPC cells was dependent on the phosphorylation status of GSK-3β.

Then, a constitutively active mutant of GSK-3β (GSK-3βS9A) was introduced into S18-shc and S26-CXCR2-CXCL5 cells. The expected enhanced expression of GSK-3β and reduced expression of Snail are presented in Fig. 6a. An analysis of the expression of epithelial and mesenchymal markers in these cells revealed that E-cadherin expression was enhanced, whereas Vimentin was downregulated (Fig. 6a). These results indicated that the upregulation of p-GSK-3β promoted an EMT phenotypic change in the NPC cells that was triggered by the CXCL5/CXCR2 axis.

High expression levels of both CXCL5/CXCR2 together results in increased activity of the ERK1/2 signalling pathway as measured by luciferase reporter assays

The luciferase reporter assays revealed that the transcriptional activity of the ERK1/2 pathway was significantly enhanced when both members of the CXCL5/CXCR2 axis were highly expressed (Fig. 6b). A western blot analysis revealed that the S18-shc and S26-CXCR2-CXCL5 cells exhibited higher levels of ERK phosphorylation than did the other stable cells. In contrast, neither the expression of ERK2 nor total ERK1/2 was altered regardless of the expression levels of CXCL5/CXCR2 (Fig. 6c and Additional file 7: Figure S4). Additionally, CXCL5 has been demonstrated to possess the ability to activate the PI3K-AKT signalling pathway in hepatocellular carcinomas [20] and colorectal cancer [25] cells via its receptor CXCR2. Therefore, we also evaluated the PI3K-AKT pathway in NPC cells. As illustrated in Additional file 8: Figure S5, the expression levels of CXCL5 and/or CXCR2 did not affect AKT phosphorylation in the S18 and S26 stable cell lines. These results demonstrated that CXCL5/CXCR2 triggered the activation of ERK by phosphorylation.

CXCL5/CXCR2 together induces hyperactivity of ERK1/2 signalling, which is necessary for the EMT and invasiveness of NPC cells

Western blot analysis (Fig. 6c) indicated that trametinib (an ERK1/2 inhibitor) successfully reduced the phosphorylation levels of the corresponding proteins and led to an epithelial phenotype in the NPC cells. Additionally, another ERK inhibitor, U0126, proved the effect of the phosphorylation of activated ERK on GSK-3β (Additional file 9: Figure S6). Moreover, the invasive activities of these cells were inhibited by trametinib (Fig. 6d). To determine whether the migration inhibitory effect of trametinib on NPC cells was associated with cell death, flow cytometry analyses were performed on S18-shc and S26-CXCR2-CXCL5 cells that were treated with 50 nM trametinib (the same concentration used in the migration assay) for 24 h. As depicted in Additional file 10: Figure S7, trametinib did not induce apoptosis in NPC cells. These results suggested that ERK1/2 signalling is necessary for the EMT and the invasiveness of the NPC cells induced by the CXCL5/CXCR2 axis.

The CXCL5/CXCR2 axis contributes to EMT by activating the ERK/GSK-3β/snail pathway in NPC cells

The results presented above suggest that the ERK/GSK-3β/Snail cascade is initiated by the CXCL5/CXCR2 axis. To further verify this suggestion, we compared the expression levels of these proteins in all of the experimental cells. As shown in Fig. 6e and Additional file 7: Figure S4, in the NPC cells that coexpressed high levels of CXCR2 and CXCL5 (i.e., S18-shc and S26-CXCR2-CXCL5 cells), the levels of p-ERK1/2, p-GSK-3β, and Snail were enhanced, and we observed the typical EMT phenotype, which included the downregulation of E-cadherin and the upregulation of Vimentin. When the S18-shc or S26-CXCR2-CXCL5 cells were treated with trametinib (an ERK1/2 inhibitor), the phosphorylation level of GSK-3β (Ser9) and the expression of Snail were reduced, and these reductions were accompanied by a reversal of the EMT phenotype, whereas the CXCR2 and CXCL5 expression was unchanged. After constitutively active GSK-3β (Ser9) plasmids were transfected into S18-shc or S26-CXCR2-CXCL5 cells, the expression of Snail was markedly reduced and this reduction was accompanied by a reversal of the EMT phenotype, but the CXCR2, CXCL5, and p-ERK1/2 expression levels remained unchanged. When siRNA interfered with the expression of Snail in S18-shc or S26-CXCR2-CXCL5 cells, we observed an upregulation of E-cadherin and a downregulation of Vimentin, but none of the aforementioned signalling molecules showed altered levels. These results further confirmed that the CXCL5/CXCR2 axis induces the EMT through the activation of the ERK/GSK-3β/Snail signalling pathway (Fig. 6f).