Fractionated irradiation-induced EMT-like phenotype conferred radioresistance in esophageal squamous cell carcinoma.

The efficacy of radiotherapy, one major treatment modality for esophageal squamous cell carcinoma (ESCC) is severely attenuated by radioresistance. Epithelial-to-mesenchymal transition (EMT) is a cellular process that determines therapy response and tumor progression. However, whether EMT is induced by ionizing radiation and involved in tumor radioresistance has been less studied in ESCC. Using multiple fractionated irradiation, the radioresistant esophageal squamous cancer cell line KYSE-150R had been established from its parental cell line KYSE-150. We found KYSE-150R displayed a significant EMT phenotype with an elongated spindle shape and down-regulated epithelial marker E-cadherin and up-regulated mesenchymal marker N-cadherin in comparison with KYSE-150. Furthermore, KYSE-150R also possessed some stemness-like properties characterized by density-dependent growth promotion and strong capability for sphere formation and tumorigenesis in NOD-SCID mice. Mechanical studies have revealed that WISP1, a secreted matricellular protein, is highly expressed in KYSE-150R and mediates EMT-associated radioresistance both in ESCC cells and in xenograft tumor models. Moreover, WISP1 has been demonstrated to be closely associated with the EMT phenotype observed in ESCC patients and to be an independent prognosis factor of ESCC patients treated with radiotherapy. Our study highlighted WISP1 as an attractive target to reverse EMT-associated radioresistance in ESCC and can be used as an independent prognostic factor of patients treated with radiotherapy.

INTRODUCTION

Radiotherapy is used as the primary curative modality for esophageal squamous cell carcinoma (ESCC), one of the most common malignancies with high morbidity and mortality in the world. However, due to radioresistance, many ESCC patients benefit little from radiotherapy, with resultant tumor recurrence or metastasis [1]. Epithelial-to-mesenchymal transition (EMT) is a cellular process in which the expressions of epithelial markers such as E-cadherin, β-catenin and occludin decrease while the expressions of mesenchymal markers including N-cadherin, vimentin and fibronectin increase [2]. Furthermore, epithelial marker β-catenin translocates from the cell membrane to the nucleus during the EMT process. Several lines of evidence have demonstrated that EMT impacts therapy response and tumor progression via multiple mechanisms [3]. Furthermore, cancer cells that have undergone EMT always display some stemness-like properties, which have been recognized as an important mediator of cell resistance to death stimuli [4]. In addition to chemotherapeutic drugs, hypoxia, microRNAs and so on [5-7], ionizing radiation has also been demonstrated to cause EMT-associated changes in some human cancers [8, 9]. In these cancers, irradiation-induced EMT was discovered to be associated with cancer cell migration and invasion. However, knowledge of irradiation-induced EMT and its exact pathological roles in ESCC is relatively scarce.

WISP1 is a secreted matricellular protein allocated to the CCN family that contains six members with highly conserved structures [10]. Recently, the CCN family has received increasing attention owing to their close relationship with EMT. For instance, Cyr61, CTGF and NOV have been reported to be associated with EMT in some human cancers [11-13]. WISP1 has been reported to play trigger roles in human idiopathic pulmonary fibrosis by regulating the expressions of EMT-associated genes. Moreover, WISP1 has been demonstrated to enhance chondrosarcoma cell motility by increasing the expression of MMP-2—one matrix metalloproteinases that facilitate the EMT event [14]. Based on the above background, we hypothesized that WISP1 may also influence the EMT process as other CCN family members do in human cancers. In our study, we found multiple fractionated irradiation induced esophageal squamous cancer cell line KYSE-150R to display significant EMT phenotype and concomitant stemness-like properties. Furthermore, WISP1 was found to be a critical mediator of EMT both in ESCC cells and in xenograft tumor models. Down-regulation of WISP1 expression in KYSE-150R significantly reversed EMT-associated radioresistance. Furthermore, WISP1 was found to be closely related to the EMT phenotype observed in ESCC patients and was an independent prognostic factor in patients treated with radiotherapy. These results show there may be important implications for the role of WISP1 in the intervention of ESCC.

MATERIALS AND METHODS

Antibodies and reagents

Antibodies against β-catenin and GAPDH were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Antibodies against E-cadherin, vimentin and N-cadherin were obtained from Epitomics (Burlingame, CA, USA). WISP1-specific neutralizing antibody α-WISP1 and recombinant WISP1 protein were purchased from Abcam (Cambridge, MA, USA). Antibodies against cleaved PARP, caspase-3, caspase-7 and caspase-9 were purchased from Cell Signaling Technology (Beverly, MA, USA).

Cell culture, human tissues and animals

The human esophageal squamous cancer cell line KYSE-150 was obtained from the Japanese Collection of Research Bioresources (JCRB, Osaka, Japan). The radioresistant esophageal cancer cell line KYSE-150R was established from KYSE-150 by multiple fractionated irradiation and had been used in our previous studies [15]. Both KYSE-150 and KYSE-150R were cultured in RPMI-1640 medium (Gibco, Life Technologies Inc., Grand Island, NY, USA) supplemented with 10% of fetal bovine serum and incubated at 37°C in 5% CO2/95% air.

The tumor specimens from ESCC patients were collected and used in our study in accordance with the guidelines of the Committees for Ethical Review of Research at Hangzhou Cancer Hospital. We collected tumor samples from 10 primary ESCC patients before radiotherapy and after radiotherapy at a total dose of 40 Gy in 20 fractions. The basic characteristics of these 10 patients are shown in Supplementary Table S1. We also collected tumor samples from another cohort of 93 primary ESCC patients to analyze the association of WISP1 expression and overall survival following radiotherapy with the basic characteristics of patients shown in Supplementary Table S2. All the tumor samples collected were diagnosed histopathologically as esophageal squamous cell carcinoma and snap-frozen in liquid nitrogen immediately after explantation until use.

Six-week-old female BALB/c nude mice or NOD-SCID mice were purchased from Vital River (Beijing, China) and maintained under standard conditions in the Experimental Animal Centre in Zhejiang Chinese Medicine University. All of the animal protocols in our study were in accordance with the institutional animal welfare guidelines of Zhejiang Chinese Medicine University.

qRT-PCR analysis

Total RNA was extracted from each cell line or tumor specimen using Trizol Reagent (Invitrogen Life Technologies, Carlsbad, CA, USA) following the manufacturer's instructions. Reverse transcription was performed with Fermentas K1622 following the manufacturer's instructions. Quantitative RT-PCR was conducted using SYBR green (Abgene, Epsom, UK) according to the manufacturer's instructions. For use of RT-PCR primers, see Supplementary Table S3.

Western blotting analysis

Protein expression was analyzed by western blotting according to the method described by Meihua Sui et al. [16]. Equal amounts of proteins (40 μg/lane) were fractionated on 12% SDS-PAGE gel and transferred to polyvinylidence difluoride membranes. The membranes were incubated with the indicated primary and secondary antibodies. Proteins were ultimately visualized using enhanced chemiluminescence and autoradiography (ECL; Thermon Scientific, Waltham, MA, USA).

Immunofluorescence analysis

For immunofluorescence analysis, KYSE-150 and KYSE-150R cells were plated on chamber slides, fixed with acetone/methanol (1:1) and then permeabilized with 0.1% Triton-X100 in PBS. Nonspecific binding sites were blocked with 3% (m/v) bovine serum albumin (BSA) in phosphate buffered saline (PBS). The cells were then incubated with indicated primary antibodies for 60 min in PBS containing 0.1% (m/v) BSA. Indirect immunofluorescence was performed by incubation with FITC or PE-conjugated secondary antibodies (Zymed; Invitrogen). Cells were ultimately mounted in vectashield mountant containing nuclei dye DAPI (Roche Diagnostics, CLSM, Nikon-A1 system, Japan). Immunofluorescence images were taken using a confocal laser scanning microscope.

Clonogenic survival assay

Exponentially growing KYSE-150 and KYSE-150R cells were seeded into six-well plates with six replicates for each group. After 24 h, adhesive cells were exposed to 2 μg/ml of WISP1 or 4 μg/ml of α-WISP1. Twenty-four hours later, cells were irradiated at various doses at an average dose rate of 300 cGy/min. Untreated cells were used as a control. Irradiated cells were cultured for another 10 days at 37°C in a 5% CO2 environment to allow colony formation. Only colonies containing ≥50 cells were counted as clonogenic survivors.

Establishment of WISP1-overexpressed KYSE-150 by lentivirus transduction

To stably up-regulate WISP1 expression, KYSE-150 was transduced with lentivirus vector containing WISP1 cDNA. Empty vector was transduced into KYSE-150 as a control. After 24 h of transduction, the lentivirus-containing medium was replaced with fresh medium and cells that were successfully transduced with lentivirus were selected with 1 ug/ml of puromycin in medium, and WISP1 up-regulation was confirmed by western blotting.

Sphere formation assay

Sphere culture was performed as previously reported [17]. Briefly, KYSE-150 and KYSE-150R cells were seeded into six-well ultra-low attachment plates (Corning, NY, USA) (1 × 103 cells/well) in serum-free DMEM/F12 medium supplemented with 20 ng/ml EGF, 5 mg/ml insulin, 0.5 mg/ml hydrocortisone, 2% B27 supplement W/O Vitamin A and 1% N2 Supplement (Invitrogen). After 15 days, only spheres with diameter ≥100 μm were counted under the microscope.

Xenograft transplantation and therapy

To develop xenograft tumors, in vitro growing cells were harvested by exposure to trypsin-ethylene diamine tetraacetic acid, washed with ice-cold PBS and implanted into the right flanks of female BALB/c nude mice (1.0 × 105 cells). When xenograft tumors had reached a mean diameter of around 0.5 cm, mice were randomly assigned into different groups (five mice in each group) and treated with PBS or radiation at a total dose of 12 Gy in three fractions every 3 days. Tumor volume (mm3) was calculated using the following formula: V(mm3) = A(mm) × B(mm)2/2, where A and B were the longest and widest diameter of tumor, respectively, and measured every 2 days with a caliper.

Immunohistochemistry analysis

For immunohistochemical analysis, paraffin-embedded sections of tumor specimens from ESCC patients were processed according to standard procedure [18]. The expression of E-cadherin, vimentin, N-cadherin, β-catenin and WISP1 were graded as 0, 1+, weak staining; 2+, strong staining in less than 30% of tumor cells; and 3+, strong staining in more than 30% of tumor cells. 0 and 1+ were defined as WISP1-negative; 2+ and 3+ as WISP1-positive. The slides were scored by a pathologist and two experienced researchers independently.

Statistics analysis

Data were presented as means ± SD from three independent experiments. Differences among the groups were examined by Student's t-test using SPSS (13.0). A probability level of 0.05 was chosen for statistical significance.

RESULTS

Fractionated irradiation induced mesenchymal-like phenotype in KYSE-150R

In our previous study, we had successfully established a radioresistant esophageal cancer cell line KYSE-150R from its parental cell line KYSE-150 by multiple fractionated irradiation [15]. Compared with KYSE-150, the radioresistant KYSE-150R showed significant changes in cell shape from cuboidal to an elongated spindle shape that resembled the mesenchymal-like phenotype (Fig. 1A). To determine whether the changes in cell shape resulted from EMT, we detected the expressions of epithelial marker E-cadherin and mesenchymal marker N-cadherin in KYSE-150R. The results showed that epithelial marker E-cadherin was significantly down-regulated and mesenchymal marker N-cadherin was significantly up-regulated in KYSE-150R compared with in KYSE-150 (Fig. 1B). Moreover, we found KYSE-150R also possessed increased translocation of β-catenin from the cell membrane into the nucleus when compared with KYSE-150 using immunofluorescence analysis (Fig. 1C). Thus, these changes in cell shape and EMT marker expressions proved that multiple fractionated irradiation promoted EMT, one anti-apoptotic event, in KYSE-150R.

Fig. 1.

Multiple fractionated irradiation conferred KYSE-150R mesenchymal-like phenotype. (A) Morphological image of KYSE-150 and KYSE-150R by phase-contrast microscopy (magnification: 20×). (B) Western blotting analysis of epithelial marker E-cadherin and mesenchymal marker N-cadherin in KYSE-150 and KYSE-150R. The graph shows the mean values (±SD) of relative expression of E-cadherin or N-cadherin versus GAPDH from three independent experiments. **P < 0.01, #P < 0.05, compared with KYSE-150. (C) Immunofluorescence analysis of the expression and cellular location of epithelial markers E-cadherin and β-catenin (magnification: 60×).

WISP1 mediated EMT-associated radioresistance in KYSE-150R

As previously described, the CCN family have been demonstrated to have an intimate relationship with EMT in some human cancers. In our study, we investigated whether this family also play critical roles in irradiation-induced EMT in KYSE-150R. We detected the mRNA levels of all the CCN family members including Cyr61, CTGF, NOV, WISP1, WISP2 and WISP3 in KYSE-150R and KYSE-150 cells. The results showed that the mRNA level of WISP1 was most significantly changed among the CCN family, with an expression increase of more than 12-fold in KYSE-150R cells compared with in KYSE-150 cells (Fig. 2A). Further studies have showed that WISP1 protein was also significantly up-regulated in KYSE-150R cells (Fig. 2B and Supplementary Figure S1A). Since the change in WISP1 expression was relatively more significant than the other CCN family members, we focused on whether WISP1 was involved in irradiation-induced EMT in KYSE-150R cells. When treated with 4 μg/ml of WISP1-specific neutralizing antibody α-WISP1 for 24 h, the EMT phenotype of KYSE-150R cells was significantly reversed, with epithelial marker E-cadherin up-regulated and mesenchymal marker N-cadherin down-regulated; in contrast, treatment with 2 μg/ml of recombinant WISP1 protein for 24 h conferred KYSE-150 cells some characteristics of mesenchymal-like phenotype, with decreased E-cadherin expression and increased N-cadherin expression (Fig. 2B, 2C and Supplementary Fig. S1A and S1B). Accompanied by the reversion of the EMT phenotype, the radioresistance of KYSE-150R cells was significantly attenuated at radiation doses of 4 Gy, 6 Gy and 8 Gy. Meanwhile, KYSE-150 cells displayed significant radioresistance at radiation doses of 4 Gy, 6 Gy and 8 Gy following the acquisition of the EMT-like phenotype (Fig. 2D). Furthermore, the levels of expression of apoptosis-related proteins including cleaved PARP, caspase-3, caspase-7 and caspase-9 were obviously increased in KYSE-150R cells that were pre-treated with 4 μg/ml of WISP1-specific neutralizing antibody α-WISP1 24 h before exposure to 8 Gy of radiation compared with in KYSE-150R cells without α-WISP1 pre-treatment. Meanwhile, these apoptosis-related proteins in KYSE-150 cells pre-treated with 2 μg/ml of recombinant WISP1 protein 24 h before exposure to 8 Gy of radiation expressed at an obviously lower level compared with that in KYSE-150 cells without WISP1 protein pre-treatment (Fig. 2E). These results suggested that WISP1-mediated EMT was involved in the development of radioresistance in KYSE-150R and KYSE-150 cells.

Fig. 2.

WISP1 mediated EMT-associated radioresistance in KYSE-150R. (A) The relative mRNA levels of the CCN family members including Cyr61, NOV, CTGF, WISP1, WISP2 and WISP3 were analyzed by qRT-PCR in KYSE-150R versus in KYSE-150. (B and C) Western blotting analysis of the expressions of WISP1, E-cadherin and N-cadherin in KYSE-150 and in KYSE-150R treated with or without 4 μg/ml of α-WISP1 antibody or 2 μg/ml of WISP1 protein for 24 h. GAPDH was used as a loading control. (D) Statistical analysis of the clonogenic survival of KYSE-150R and KYSE-150 receiving indicated pre-treatments before exposure to 2, 4, 6 and 8 Gy of radiation. * and #, P < 0.05, **P < 0.01, compared with KYSE-150R; &P < 0.05, &&P < 0.01, compared with KYSE-150. € Western blotting analysis of the expressions of apoptosis-related proteins including cleaved PARP, caspase-3, caspase-7 and caspase-9 24 h after 8 Gy of radiation in KYSE-150 cells with or without pre-treatment with 2 μg/ml of recombinant WISP1 protein for 24 h, and in KYSE-150R cells with or without pre-treatment with 4 μg/ml of WISP1-specific neutralizing antibody α-WISP1 for 24 h. GAPDH was used as a loading control.

WISP1 caused EMT-associated radioresistance in xenograft tumor models

To further confirm WISP1 was capable of inducing EMT in vivo, we transplanted KYSE-150 cells transduced with lentivirus vector carrying WISP1 cDNA or empty vector as a control into BALB/c nude mice to develop xenograft tumor models. The tumor-bearing mice were treated with PBS or 12 Gy of radiation (single dose of 4 Gy every 3 days). Immunohistochemistry analysis showed that epithelial markers E-cadherin and β-catenin were significantly down-regulated, while mesenchymal markers N-cadherin and vimentin were up-regulated in WISP1-overexpressed tumors compared with in control tumors (Fig. 3). Furthermore, we found irradiation induced down-regulation of E-cadherin and β-catenin and up-regulation of N-cadherin and vimentin either in control tumors or in WISP1-overexpressed tumors (Fig. 3). These results suggested that WISP1 also promoted the EMT-like process in vivo by regulation of EMT-associated genes. Since EMT-associated changes were observed in xenograft tumor models, we further investigated whether these changes led to the enhancement of radioresistance. We found the volume and weight of WISP1-overexpressed tumors were significantly higher than control tumors following radiotherapy (Fig. 4A and 4B). Immunohistochemical analysis showed that the expression of WISP1 was significantly higher in WISP1-overexpressed tumors than in control tumors either before radiotherapy or after radiotherapy (12 Gy in three fractions) (Fig. 4C). The WISP1-overexpressed tumors showed significantly increased cell proliferation compared with control tumor after treatment with 12 Gy of radiation in three fractions as determined by the staining of Ki-67, a proliferation-associated protein only present in proliferating cells (Fig. 4D and 4E) [19]. These results suggest that WISP1-induced EMT is closely related to increased radioresistance in xenograft tumor models.

Fig. 3.

WISP1 mediated EMT phenotype in xenograft tumor models. (A and B) Immunohistochemical analysis of the expressions of epithelial markers E-cadherin and β-catenin and mesenchymal markers N-cadherin and vimentin in WISP1-overexpressed xenograft tumors and in control tumors transduced with empty vector before radiotherapy and after 12 Gy of radiation in three fractions (magnification: 20×). (C) Quantitative analysis of the staining intensity scores of E-cadherin, β-catenin, N-cadherin and vimentin in ‘A’ and ‘B’ according to the method described in the ‘Materials and Methods’ section. *P < 0.05, **P < 0.01, compared with tumors transduced with empty vector; #P < 0.05, ##P < 0.01, compared with irradiated WISP1-overexpressed tumors; &P < 0.05, &&P < 0.01, compared with irradiated tumors transduced with empty vector.

Fig. 4.

WISP1 induced EMT-associated radioresistant phenotype in xenograft tumor models. (A) The growth curve of WISP1-overexpressed tumors and control tumors transduced with empty vector with or without treatment with 12 Gy of radiation in three fractions every 3 days. (B) The weight of xenograft tumors in ‘A’ at the termination of the experiment on day 25. *P < 0.05, compared with irradiated tumors transduced with empty vector. (C) The expression of WISP1 protein in xenograft tumors in ‘A’ at the termination of the experiment on day 25 by immunohistochemical analysis (magnification: 20×). (D) The staining of Ki-67, a proliferation-associated protein, in xenograft tumors in ‘A’ at the termination of the experiment on day 25 by immunohistochemical analysis (magnification: 20×). € Quantitative analysis of Ki-67 expression percentage in ‘D’ based on five randomly selected high-power microscopic fields. *P < 0.05, compared with irradiated tumors transduced with empty vector.

WISP1 was an independent prognostic factor of ESCC patients following radiotherapy

In addition to xenograft tumor models, we also examined whether EMT occurred in ESCC patients treated with radiotherapy. Using western blotting analysis of tumor specimens from 10 primary ESCC patients, epithelial marker E-cadherin was found to be significantly down-regulated, while mesenchymal marker N-cadherin was significantly up-regulated after radiotherapy at a total dose of 40 Gy in 20 fractions compared with before radiotherapy (Fig. 5A and Supplementary Fig. S2). These results suggest that ionizing radiation also promotes the EMT phenotype in ESCC patients. Interestingly, we found the expression of WISP1 significantly increased in the 10 tumor samples where the EMT phenotype was observed (Fig. 5A and Supplementary Fig. S2), suggesting that WISP1 was also closely related to EMT in ESCC patients. By immunohistochemical analysis of another cohort of 93 primary ESCC patients, both the WISP1-positive ratio and WISP1 expression intensity were demonstrated to be significantly increased after 40 Gy of radiation compared with before radiotherapy either in the WISP1-positive group or the WISP1-negative group (Fig. 5B and Supplementary Fig. S3). Furthermore, survival analyses showed that WISP1-positive patients had significantly poorer prognoses than WISP1-negative patients after radiotherapy, suggesting that WISP1 was an independent prognosis factor of ESCC patients treated with radiotherapy (Fig. 5C). These results together prove that irradiation-induced WISP1 up-regulation may be closely related to the EMT phenotype in vivo that contributes to the unfavorable prognosis of ESCC patients.

Fig. 5.

WISP1 was an independent prognostic factor of ESCC patients treated with radiotherapy. (A) Protein expressions of E-cadherin, N-cadherin and WISP1 in tumor specimens from 10 primary ESCC patients before radiotherapy and after radiotherapy at a total dose of 40 Gy in 20 fractions by western blotting analysis. GAPDH was used as a loading control. (B) One representative result of WISP1 protein expression in another cohort of 93 primary ESCC patients before radiotherapy and after radiotherapy at a total dose of 40 Gy in 20 fractions by immunohistochemical analysis (magnification: 40×). (C) Kaplan–Meier analysis showed that WISP1 expression was significantly associated with overall survival of ESCC patients treated with radiotherapy.

Acquisition of some stemness-like properties in KYSE-150R

There are several studies demonstrating that cells with the EMT phenotype always acquire some stemness-like properties [4]. In our study, we also investigated whether KYSE-150R cells with EMT phenotype possessed some stemness-like properties. Using a colony-forming assay, we found KYSE-150R cells displayed significant density-dependent growth promotion, which is considered to be one of the most important properties of stem cells in normal tissues such as bone marrow cells and neural progenitor cells [20, 21]; however, this growth pattern was not observed in KYSE-150 cells (Fig. 6A). Moreover, KYSE-150R and KYSE-150 cells were injected into immunodeficient NOD-SCID mice at cell numbers of 102, 103 or 104 to detect tumor formation ability. The results showed that 104 KYSE-150R cells obviously reproduced the original tumor in mice within 4 weeks, but the same number of KYSE-150 cells only formed small nodes in part of the mice (Fig. 6B and 6C). There was no macroscopically detectable tumorigenesis for either KYSE-150 or KYSE-150R at cell numbers of 102 or 103. Furthermore, in comparison with KYSE-150, KYSE-150R showed significantly greater capability for sphere formation, another known stemness-like property (Fig. 6D). These results indicated that KYSE-150R cells with the EMT phenotype simultaneously acquire some stemness-like properties—one recognized mechanism of chemoradiotherapy resistance.

Fig. 6.

The acquisition of stemness-like properties in KYSE-150R. (A) The determination of density-dependent growth promotion in KYSE-150R and KYSE-150 by clonogenic survival assay. (B) The tumorigenesis of KYSE-150R and KYSE-150 at cell numbers of 104 4 weeks post-implantation into NOD-SCID mice. (Insert) photos of excised xenograft tumors at the termination of the experiment. (C) The weight of excised xenograft tumors in ‘B’, *P < 0.05 compared with xenograft tumors implanted with KYSE-150R. (D) The number of spheres of KYSE-150 or KYSE-150R formed was counted under sphere culture conditions as indicated in the ‘Materials and Methods’ section. ***P < 0.001, compared with KYSE-150.

DISCUSSION

EMT is a critical cellular process where cells lose their epithelial phenotype that allows them to grow adherently and display apicobasal polarity. Furthermore, the cells undergoing EMT concomitantly acquire some mesenchymal properties characterized by fibroblast-like cell shape, elevated resistance to apoptosis following death stimulus and enhanced cell migration and invasion [22]. EMT is fundamental for appropriate embryonic development, but this cellular process is re-activated in adults during organ fibrosis, wound healing and cancer progression [23]. In our study, we found that multiple fractionated irradiation-induced esophageal squamous cancer cell line KYSE-150R cells developed EMT-like changes. In lung cancer, breast cancer, colon cancer and other cancers, irradiation-induced EMT-associated changes have been studied only in vitro without further validation in vivo [8, 9]. In our study, we further investigated whether ionizing radiation induced EMT-associated changes in ESCC patients treated with radiotherapy. We found epithelial marker E-cadherin was down-regulated, while mesenchymal marker N-cadherin was up-regulated after 40 Gy of radiation in 20 fractions compared with before radiotherapy in ESCC patients. This implied that ionizing radiation mediated EMT, an anti-apoptotic program, both in vitro and in vivo, greatly attenuating the radiation effect of ESCC treatment.

Furthermore, the molecular determinant of fractionated irradiation-induced EMT was dissected in KYSE-150R cells. WISP1, one member of the CCN family, was found to be a critical mediator of EMT both in ESCC cells and in xenograft tumor models. Members of the CCN family with highly conserved structures were expressed in several human cancers and were closely associated with EMT as a positive or negative regulator. For instance, CTGF was capable of inducing EMT-associated changes in human peritoneal mesothelial cells, whereas WISP2 and WISP3 inhibited EMT signaling pathways [11, 13, 24]. However, whether WISP1 influences EMT in human cancers has not yet been reported. Here, our findings reveal WISP1 to be a mediator of EMT in ESCC. Previously, WISP1 has been reported to be highly expressed in esophageal cancer tissues compared with in adjacent benign esophageal tissues, with an adverse effect on prognosis [25]. However, the precise mechanisms by which high expression of WISP1 resulted in poor prognosis of ESCC patients have remained poorly elucidated. Our findings have at least partially revealed how WISP1 induces an unfavorable prognosis in patients with ESCC.

Several lines of evidence have suggested that cells that have undergone EMT are likely to show some stemness-like properties [26]. In our study, the esophageal cancer cell line KYSE-150R with the EMT-like phenotype showed relative quiescence and strong capability for spherical growth and tumorigenesis, both of which were suggestive of stemness-like properties [27]. EMT is closely related to stemness in breast cancer, lung cancer and other cancers [28, 29]. Here, our studies have shown that esophageal cancer cells also acquire some stemness-like properties concomitantly with the acquisition of the EMT phenotype. In a previous study by Guanyu Wang et al., Bmi-1 was demonstrated to be significantly up-regulated in KYSE-150R cells compared with in their parental KYSE-150 cells [30]. Mounting evidence has shown that Bmi-1 is involved in regulation of self-renewal and differentiation of stem cells via multiple mechanisms. For instance, Bmi-1 has been shown to be capable of inducing telomerase activity and down-regulating p16INK4a and p19ARF expression, allowing cells to bypass senescence and to be immortalized [31]. In our study, we proposed that the acquisition of stemness-like properties was possibly due to up-regulation of Bmi-1 expression. However, whether the expression of Bmi-1 is regulated by WISP1 needs to be further investigated in future work. Since stemness is a recognized cell survival mechanism through preferential activation of the DNA damage response following ionizing radiation [32], those stemness-like properties observed in KYSE-150R are at least partially responsible for EMT-associated radioresistance. In conclusion, our study highlighted WISP1 as an attractive target to reverse EMT-associated radioresistance in ESCC. However, more efforts are still needed to clarify the precise mechanism by which EMT gives rise to radioresistance in ESCC.

FUNDING

Funding to pay the Open Access publication charges for this article was provided by the (No. 81402529), (No. LY14H160005 and No. LZ15H220001), (No. 2012ZD002 and No. 2015Z04), (No. 20150733Q64) and (No. 2015KYB325 and No. 2015PYA009).