Hes1 triggers epithelial-mesenchymal transition (EMT)-like cellular marker alterations and promotes invasion and metastasis of nasopharyngeal carcinoma by activating the PTEN/AKT pathway.

Overexpression of the transcriptional factor Hes1 (hairy and enhancer of split-1) has been observed in numerous cancers, but the precise roles of Hes1 in epithelial-mesenchymal transition (EMT), cancer invasion and metastasis remain unknown. Our current study firstly revealed that Hes1 upregulation in a cohort of human nasopharyngeal carcinoma (NPC) biopsies is significantly associated with the EMT, invasive and metastatic phenotypes of cancer. In the present study, we found that Hes1 overexpression triggered EMT-like cellular marker alterations of NPC cells, whereas knockdown of Hes1 through shRNA reversed the EMT-like phenotypes, as strongly supported by Hes1-mediated EMT in NPC clinical specimens described above. Gain-of-function and loss-of-function experiments demonstrated that Hes1 promoted the migration and invasion of NPC cells in vitro. In addition, exogenous expression of Hes1 significantly enhanced the metastatic ability of NPC cells in vivo. Chromatin immunoprecipitation (ChIP) assays showed that Hes1 inhibited PTEN expression in NPC cells through binding to PTEN promoter region. Increased Hes1 expression and decreased PTEN expression were also observed in a cohort of NPC biopsies. Additional studies demonstrated that Hes1-induced EMT-like molecular changes and increased motility and invasion of NPC cells were mediated by PTEN. Taken together, our results suggest, for what we believe is the first time, that Hes1 plays an important role in the invasion and metastasis of NPC through inhibiting PTEN expression to trigger EMT-like phenotypes.

It has been reported that Hes1 showed elevated expression in the metastatic PC-3 and PC-3M human prostate cancer cells [34]. In a preliminary study, we also observed that Hes1 expression correlated with lymph node metastasis of NPC (Supplementary Table S2). Briefly, the expression of Hes1 was higher in NPC tumors at the N2-N3 and M1 stages than tumors at the N0-N1 and M0 stage, respectively. These observations suggest that Hes1 is involved in cancer invasion and metastasis. However, the underlying mechanisms are not clear.
Invasion and metastasis are significantly aggres sive phenotypes of human cancers and the most lethal attributes of cancer deaths. Prevention of tumor invasion and metastasis has attracted great attention in clinical research [35][36][37][38][39]. Previous studies suggest that epithelialmesenchymal transition (EMT) is a central mechanism contributing to the invasion and metastasis of various cancers [36,40]. Therefore, identification of key factors involved in EMT and investigation of the molecular mechanism of EMT are of critically importance for understanding tumor invasion and metastasis and developing novel interventions for metastatic cancers. In the present study, we investigate whether Hes1 is involved in EMT, and invasion and metastasis of NPC cells, as well as the underlying mechanisms. Our results showed, for the first time, that Hes1 triggered EMTlike cellular marker alterations and promoted invasion and metastasis of NPC by activating the PTEN/AKT pathway.

Hes1 overexpression was frequently detected in NPC biopsies and cell lines
We first evaluated the expression of Hes1 protein in 103 paraffinembedded, archived NPC biopsies and 29 paraffinembedded, archived non cancerous nasopharyngeal epithelial biopsies using immunohistochemistry (IHC) staining. High expression of Hes1 was detected in 59 of 103 NPC specimens (57.3%) and 10 of 29 non-cancerous nasopharyngeal epithelial tissues (34.5%), respectively (Figure 1A, 1B; Supplementary Table S1). Therefore, Hes1 overexpression was more frequently detected in NPC than non-cancerous nasopharyngeal epithelial biopsies.
In addition, the localization of Hes1 was found in both the nucleus and cytoplasm of NP69, CNE2, and SUNE1 cells, based on immunofluorescence assay (Supplementary Figure S1), and the distribution of Hes1 was detected in the nucleus and cytoplasm of cancer cells based on IHC ( Figure 1A, 1C).

Hes1 overexpression was associated with aggressive and/or poor prognostic phenotypes of NPC
The relationship between Hes1 expression and a number of clinicopathologic characteristics of NPC patients was summarized in Table 1. No significant association was identified between Hes1 expression and the age (P = 0.162), sex (P = 0.882), histological subtype (P = 0.079), and NPC recurrence (P = 0.718) of 103 NPC cases (Table 1). On the contrary, Hes1 expression was associated with the tumor size (T classification) (P = 0.018), lymph node invasion (N classification) (P = 0.006), metastasis (M classification) (P = 0.021), and clinical stage (P = 0.001) of 103 NPC patients (Table 1). Briefly, high expression of Hes1 was more frequently observed in T3-T4, N2-N3, M1, and III-IV tumors than  T1-T2, N0-N1, M0, and I-II tumors, suggesting that Hes1 overexpression correlated with aggressive phenotypes of NPC and may be involved in the invasion and metastasis of NPC ( Figure 1C, 1D and Table 1).
Given that Hes1 overexpression was associated with advanced NPC, we further evaluated the prognostic value of Hes1 expression for NPC patients. A negative correlation between the level of Hes1 protein expression and the overall survival of NPC patients was identified based on Kaplan-Meier analysis of the log-rank test ( Figure 1E and Supplementary Table S2). The patients with higher levels of Hes1 expression had poorer overall survival than those with lower levels of Hes1 expression (P = 0.002; Figure 1E and Supplementary  Table S2). Moreover, N/M classifications and clinical Low expression of Hes1 in noncancerous nasopharyngeal biopsies; c. and d. High expression of Hes1 in noncancerous nasopharyngeal biopsies; e. and f. Low expression of Hes1 in NPC biopsies; g. and h. High expression of Hes1 in NPC biopsies. The brown staining indicates Hes1 immunoreactivity. B. Hes1 expression was significantly higher in the NPC biopsies than that in the noncancerous nasopharyngeal biopsies. C. Representative images of Hes1 expression in NPC biopsies of different TNM stages. High expression of Hes1 was observed in the T4 (b), N3 (d), M1 (f) and IV (h) stages NPC biopsies, while low expression of Hes1 was detected in the T1 (a), N0 (c), M0 (e) and I (g) stages of tumor. D. The numbers and percentages of high and low expression of Hes1 according to different clinicopathological features. E. Cumulative overall survival curves of 103 NPC patients with different Hes1 expression (high or low). The patients with high Hes1 expression had a significantly shorter overall survival than the patients with low Hes1 expression. The P values were calculated by log-rank test. www.impactjournals.com/oncotarget stages were also significantly associated with the overall survival of NPC patients (P < 0.05, all) (Supplementary  Table S2). Multivariate analysis was conducted to identify independent prognostic factors for NPC patients. Multivariate Cox regression analysis showed that distant metastasis (P = 0.011) and tumor recurrence (P = 0.000) were independent prognostic factors for NPC patients, whereas Hes1 expression was not an independent prognostic factor for NPC patients (P = 0.053) (Supplementary Table S2). Thus, Hes1 overexpression was significantly associated with poor prognosis of NPC patients, whereas Hes1 expression was not an independent prognostic factor for NPC patients.

Ectopic expression of Hes1 in NPC cells induced EMT-like molecular changes and enhanced cell motility and invasion
As mentioned above, the data from clinical NPC specimens showed that Hes1 overexpression correlated with the invasive and metastatic properties of NPC. EMT is a central mechanism contributing to invasion and metastasis of various cancers [36,40]. To understand whether Hes1 overexpression directly induces EMT and invasion and motility of NPC cells, we examined the surface markers and phenotypic changes of NPC cells with ectopic expression of Hes1. The Hes1 transgene was  Figure 2A, 2B, 2C). The qRT-PCR results demonstrated ectopic expression of Hes1 significantly reduced the expression of epithelial markers (i.e., E-cadherin and βcatenin) and significantly increased the expression of mesenchymal markers (vimentin and N-cadherin) in CNE2 and SUNE1 cells (Figure 2A, 2B). In addition, Western blotting results also revealed that Hes1expressing CNE2 and SUNE1 cells exhibited typical EMT-like phenotypes, including downregulation of epithelial markers Ecadherin, αcatenin, and βcatenin, and upregulation of mesenchymal markers vimentin, fibronectin, and Ncadherin ( Figure 2C). EMT is one of the most important steps in tumor metastasis and advance by improving the motility and invasion of tumor cells [36,40]. Therefore, we also examined the effects of Hes1 on the motility and invasion of NPC cells based on transwell migration and Boyden invasion assays. As shown in Figure 2D, Hes1-expressing CNE2 and SUNE1 cells displayed significantly enhanced mobility and invasion abilities in comparison with vectorexpressing cells. Taken together, these results suggest that Hes1 overexpression enhanced the mobility and invasion of NPC cells in vitro by inducing EMT-like cellular marker alterations.

Silence of endogenous Hes1 reversed EMT-like phenotypes and reduced the migration and invasion abilities of NPC cells
To further examine the effects of Hes1 on EMT, migration, and invasion of NPC cells, endogenous Hes1 in CNE2 and SUNE1 cells was silenced using specific shRNA and the phenotypes were compared with wild-type NPC cells. The shRNAHes1 specifically knocked down endogenous Hes1 mRNA ( Figure 3A, 3B) and protein ( Figure 3C) expression in both CNE2 and SUNE1 cells. As indicated in Figure 3A, 3B, 3C, silencing endogenous Hes1 in CNE2 and SUNE1 cells increased the expression of epithelial markers (i.e., Ecadherin, αcatenin, and βcatenin) and concomitantly reduced the expression of mesenchymal markers (i.e., vimentin, fibronectin, and N-cadherin) at both mRNA and protein levels. Transwell migration and Boyden invasion assays also demonstrated that knockdown of endogenous Hes1 by shRNA markedly inhibited the migration and invasion of CNE2 and SUNE1 cells ( Figure 3D). Taken together, suppression of endogenous Hes1 expression in NPC cells reversed EMTlike molecular changes and reduced the migration and invasion of NPC cells.

Hes1 positively modulates the metastasis of NPC cells in vivo
Local invasion and distant metastasis are common clinical features of NPC. To determine whether Hes1 improves the migration and invasion of NPC cells in vivo, the vector-or Hes1-expressing CNE2 cells were subcapsularly transplanted into the liver of nude mice and the formed tumors in liver and potentially metastatic tumors on lung surface were evaluated. We found that the size of tumors on liver surface derived from Hes1expressing CNE2 cells was significantly larger than that derived from vector-expressing CNE2 cells ( Figure 4A). No intrahepatic metastasis of tumors was detected in mice inoculated with vector-or Hes1-expressing CNE2 cells (data not shown). Moreover, IHC analysis revealed that Hes1 expression of primary tumors derived from Hes1expressing CNE2 cells was significantly higher than that of primary tumors derived from vecto-expressing CNE2 cell ( Figure 4B).
While metastatic tumors on lung surface derived from vector-or Hes1-expressing CNE2 cells were not detected under a dissection microscope ( Figure 4C), histological examination identified an average of 23.3 ± 23.1 metastatic tumor nodules on lung surface in three of eight mice received CNE2-vector transplant ( Figure 4D, 4E, 4F). In contrast, an average of 61.4 ± 46.7 metastatic tumor nodules were detected on lung surface in seven of eight mice inoculated with Hes1-expressing cells ( Figure 4D, 4E, 4F) based on histological examination. Furthermore, H&E staining showed that the size of micrometastatic lesions was markedly larger in the lungs of mice injected with Hes1-expressing cells than in the mice inoculated with CNE2-vector ( Figure 4D). In addition, RNAi-mediated silencing of endogenous Hes1 in CNE2 cells significantly inhibited the metastatic ability of NPC cells in vivo (Supplementary Figure S2). Thus, Hes1 positively regulates the metastasis of NPC cells in vivo.

Hes1 inhibited the expression of PTEN by binding to its promoter region
As a transcriptional repressor [17,41,42], Hes1 is not able to directly activate the PI3K/Akt pathway. It was speculated that Hes1-dependent activation of the PI3K/ Akt pathway was mediated by transcriptional repression of an AKT-negative regulator through Hes1. It is well known that the tumor suppressor PTEN negatively regulates the PI3K/Akt pathway [43][44][45]. Therefore, we compared PTEN expression among Hes1-expressing, shHes1expressing, and relative control cells. The qRT-PCR and Western blotting results revealed that Hes1 overexpression led to downregulation of PTEN mRNA and protein levels, while Hes1 silence resulted in a significant upregulation of PTEN at both mRNA and protein levels ( Figure  5A, 5B), suggesting that Hes1 negatively regulates PTEN expression. We next performed chromatin immunoprecipitation (ChIP) assay to examine whether Hes1 is associated with the PTEN locus. ChIP analysis showed that Hes1 bound to the regulatory sequences in the PTEN promoter region in CNE2 cells ( Figure 5C), which was validated by quantitative ChIP assays ( Figure 5D). The promoter luciferase reporter assays revealed that the DNA sequence has promoter activity and Hes-1 directly regulates the promoter activity of PTEN [46]. Taken together, these results suggest that Hes1 transcriptionally inhibited the expression of PTEN by binding to its promoter region.

Hes1 induced EMT-like molecular changes and enhanced migration and invasion of NPC cells through inhibiting the expression of PTEN
To analyze the functional correlation between PTEN and Hes1, we further examine whether stable shRNA-mediated knockdown of PTEN expression in Hes1-silenced CNE2 cells could reactivate PI3K/ AKT/GSK3β signaling pathway. As shown in Figure  5E, PTEN overexpression markedly reduced AKT and GSK3β phosphorylation in CNE2 cells, similar as the results caused by Hes1 silencing. In addition, silencing endogenous PTEN with PTEN-shRNA in Hes1-silenced CNE2 cells increased the levels of phosphorylated AKT (Ser473) and GSK3β, and snail ( Figure 5E).
To understand whether Hes1-induced EMT and enhanced motility and invasion of NPC are mediated by PTEN, we performed gain-of-function and loss-of-function experiments. First, we examined the role of PTEN in EMT, and migration and invasion of NPC cells by re-expression of PTEN. As shown in Figure 5, gain of PTEN function through exogenous expression of PTEN increased the expression of epithelial markers (E-cadherin and αcatenin) ( Figure 5E), reduced the expression of mesenchymal markers (vimentin, fibronectin, Ncadherin and snail) ( Figure 5E), and inhibited the motility and invasion of CNE2 cells, which were similar as the results caused by Hes1 silencing ( Figure 5F).  Subsequently, we investigated whether ectopic expression of shPTEN reverses shHes1-induced mesenchymal-epithelial transition (MET) and rescues shHes1-mediated inhibition of migration and invasion of CNE2 cells. We found that shPTEN reversed METlike phenotypes ( Figure 5E) and abrogated the inhibition  Figure 5F) induced by shHes1. Furthermore, shRNA-mediated PTEN knockdown enhanced migration and invasion of CNE2 cells, similar to the results of ectopic expression of HES1 in NPC cells ( Figure 5F). PTEN overexpression in Hes1-expressing cells abrogated the increased motility and invasion of CNE2 cells ( Figure 5F) induced by enforced expression of Hes1. Taken together, these results suggest that PTEN is involved in Hes1-induced EMT-like phenotypes, motility, and invasion of NPC cells.

Association among Hes1 and PTEN expression and EMT phenotypes in NPC patients
As mentioned above, Hes1 expression was increased in NPC biopsies ( Figure 1A, 1B; Supplementary Table S1). Our study (Supplementary Table 3) and previous studies [47,48] observed reduced expression of PTEN in NPC tissues [47][48][49]. Therefore, we used IHC to identify any association between the expression of PETN and Hes1 in NPC biopsies. The expression level of Hes1 and PTEN was analyzed in 103 cases of human primary NPC tissues based on IHC. We found that a negative association between PTEN and Hes1 expression in the 103 NPC biopsies ( Figure 6B, 6C and Supplementary Table S4).
Furthermore, high expression of Hes1 is associated with high expression of fibronectin and snail, and low expression of E-cadherin in the NPC biopsies ( Figure  6A, 6C and Supplementary Table S4). Statistical analyses revealed a significant negative association between Ecadherin and Hes1 expression, and a significant positive association between both fibronectin and snail and Hes1 expression in NPC biopsies ( Figure 6A, 6C and Supplementary Table S4). These results suggest that high expression of fibronectin and snail and low expression of PTEN and Ecadherin were significantly associated with Hes1 overexpression in NPC cells undergoing EMT, invasion, and metastasis. Taken together, our results revealed a negative association between high expression of Hes1/fibronectin/snail and low expression of PTEN/E cadherin in NPC cells, supporting a model in which Hes1 transcriptionally inhibits PTEN expression to activate the PI3K/Akt/GSK3β/snail signaling pathway and downregulate E-cadherin (Figure 7).

DISCUSSION
Although double immunofluorescence have shown that Hes1, a downstream target of Notch1 signaling, was expressed in Sox2-and Oct4-positive cells in human primary NPC tissue [50], the present study was the first one demonstrating that the expression of Hes1 was significantly increased in NPC biopsies compared to normal tissues. Increased levels of Hes1 mRNA and/ or protein have also been observed in a number of solid cancers [20][21][22][23][24][25][26][27][28][29][30]. To our knowledge, the present study was also the first one reporting that Hes1 expression is associated with the invasion and metastasis of human NPC, such as the tumor size (T classification), lymph node metastasis (N classification), and distant metastasis (M classification), which has never been reported in NPC and other cancers. Previous studies have reported that Hes1 overexpression was associated with poor survival of patients with glioma [25] and medulloblastomas [30]. Results of these studies and the present study suggest that Hes1 overexpression is a common feature in human cancers and the expression of Hes1 may be served as a helpful prognostic marker to evaluate the prognosis of NPC patients.
In the present study, we observed Hes1 expression in both cytoplasm and nuclei of malignant and nonmalignant cells and tissues, which is consistent with the previous reports in colorectal cancer [20], pancreatic cancer [31], and rhabdomyosarcoma [51]. These results suggest that Hes1 was synthesized in cytoplasm and transported into nuclei to function as a transcriptional repressor. However, the dynamic distribution and biological activities of Hes1 in the development of NPC need to be further investigated.
Previous study reported that the expression of Hes1 in C4-2B cells, a highly metastatic derivative of the LNCaP cell line, was 20-fold higher than that in LNCaP cells [34]. In our study, we identified the association of high expression of Hes1 and the invasion and metastasis in 103 NPC cases. Gain-of-function and loss-of-function studies demonstrated that high expression of Hes1 was associated with improved migration and invasion of NPC cells in vitro. In addition, exogenous expression of Hes1 led to invasion and metastasis of NPC cells in mouse NPC model. These results strongly support the role of Hes1 in invasion and metastasis of NPC. The previous studies reported an association between Bmi-1 expression and local invasion of primary tumor of NPC. However, no association between Bmi-1 expression and metastasis was identified in NPC biopsies, suggesting that overexpression of Bmi1 alone is not sufficient to induce metastasis of NPC, and Bmi-1 is mainly involved in local invasion of tumors [45,52]. In addition, it has been shown that Bmi-1 overexpression trigged EMT to improve the invasion property of NPC cells [45,52]. In the present study, our results suggest that overexpression of Hes1 alone is sufficient to induce local invasion and metastasis of NPC. Therefore, Hes1 may play a more important role in the invasion and metastasis of NPC than Bmi-1.
The majority of cancer deaths are due to tumor metastasis, therefore, metastasis prevention of tumor cells has received much attention [35][36][37][38][39]. Understanding the molecular mechanisms underlying cancer invasion and metastasis may lead to novel therapeutic approaches against cancers [35][36][37][38][39]. EMT is an early and characteristic event in the invasion and metastasis of various cancers www.impactjournals.com/oncotarget [36,40]. In the present study, we found that upregulation of Hes1 triggered EMT-like cellular marker alterations and enhanced the motility and invasion of NPC cells, whereas silencing endogenous Hes1 expression reversed EMTlike phenotypes and reduced cell motility and invasion. Therefore, Hes1 may be an important target for the prevention of invasion and metastasis of NPC.
In our previous studies, expression profiling of various biomarkers involved in EMT was performed using tissue microarray-based immunohistochemistry in 122 clinical specimens of NPC [53,54]. Our results showed that neoplastic spindle cells in the clinical specimens of NPC exhibited EMT features [54]. In addition, we found that up-regulation of EMT markers (vimentin, fibronectin, MMP2, periostin, SPARC, snail, and slug), and E-to N-cadherin switch, occurred preferentially in tumors containing a large proportion of spindle-shaped malignant cells [53,54]. Therefore, in the present study, we investigated the expression of Hes1 and EMT biomarkers in a larger cohort of 103 NPC samples from the previous studies [53,54] to understand the clinical correlation between Hes1 expression and EMT biomarkers. Our results demonstrated that the high expression of fibronectin and snail and the low expression of Ecadherin were significantly associated with the gain of Hes1 expression in EMT, invasion and metastasis of NPC, suggesting that Hes1 plays an important role in triggering EMT in NPC specimens, which was consistent with the results of in vitro studies described above. To the best of our knowledge, this is the first study identifying the essential role of Hes1 in induction of EMT-like phenotypes, invasion and metastasis of cancers. It is well known that cells undergoing EMT change from a cobblestone-like appearance to an elongated and mesenchymal phenotype [35][36][37][38][39]. The previous report showed that although no EMT-associated morphological changes were observed in Bmi-1-overexpressing immortalized human mammary epithelial cells (HMECs), Bmi-1 overexpression in immortalized HMECs repressed epithelial markers and up-regulated mesenchymal markers, and promoted cellular motility and invasion [55]. Conversely, the repression of Bmi-1 reversed the expression of EMT markers [55]. It is very clearly that Bmi1 alone is not sufficient to induce the typical EMT morphological changes in immortalized HMECs. Furthermore, the morphologic changes of EMT might be directed by differential oncogene activation. Ras [56] and ILEI [57] can lead to EMT, tumor formation and metastasis. These findings suggest that additional oncogenic events, such as H-Ras expression or loss of expression of tumor suppressor genes could be involved in the EMT of immortalized HMECs induced by Bmi-1. Thus, these authors [55] thought that Bmi-1-induced EMT is celltype specific. In this study, Hes1 overexpression didn't induce the typical EMT morphological changes in NPC cells, but increased the motility and invasive properties of NPC cells, which is concurrent with the increased expression of mesenchymal markers and the decreased expression of epithelial markers. More importantly, high expression of Hes1 is associated with high expression of mesenchymal markers (fibronectin and snail), and low expression of E-cadherin in the NPC biopsies. Together, Hes1 alone is not sufficient to trigger the typical EMT morphological changes (i.e., mesenchymal-like morphological conversion) in NPC cell types (CNE2 and SUNE1 cells) examined.
βcatenin is frequently used as a mesenchymal marker gene as a target of WNT pathway. In our study, Western blot analysis and qRT-PCR illustrated that ectopic expression of Hes1 in NPC cells significantly reduced βcatenin expression ( Figure 2). Our observations revealed that Hes1 overexpression causes the reduced expression of membranous and cytoplasmic βcatenin, and didn't lead to nuclear βcatenin accumulation in tumor section formed from Hes1-expressing CNE2 cells (Supplementary Figure S3). In NPC, βcatenin is used as both mesenchymal marker gene [58][59][60][61] and epithelial marker gene [62,63]. Thus, in this case, we suggest that βcatenin can be used as an epithelial marker involved in Hes1-induced EMT-like cellular marker alterations.
Subsequently, we identified PTEN as a potential target of Hes1. It has been reported that loss of function or downregulation of PTEN was associated with metastasis, more aggressive growth behaviors, and/or poor prognostic phenotypes of many cancers including NPC [45,48,[64][65][66][67][68][69][70][71][72][73]. In the present study, we provided evidence showing that PTEN was a potentially functional target of Hes1 and involved in the metastasis and invasion of NPC. First, a negative correlation between Hes1 and PTEN expression was observed in NPC cells and specimens. Second, suppression of PTEN expression by shRNA in Hes1-repressed CNE2 cells rescued PI3K/AKT activities and the migration/invasion of cancer cells. In addition, we found that Hes1 bound to the promoter region of PTEN. Moreover, the promoter luciferase reporter assays revealed that the DNA sequence has promoter activity and Hes-1 directly regulates the promoter activity of PTEN [46]. Collectively, these results suggest that PTEN is a bona fide target of Hes1, which negatively modulates PTEN expression.
Next, we investigated whether Hes1 triggered EMT, migration and invasion of NPC cells in vitro through downregulating PTEN. It has been reported that PTEN was involved in the regulation of EMT during embryonic development and cancer progress by downregulating the PI3K/Akt pathway [45]. Increasing evidence supports that inactivation or downregulation of the tumor suppressor PTEN triggers EMT of cancer  [45,[74][75][76], which then promotes the invasion and metastasis of various cancers including NPC [45,75]. In the present study, we found that PTEN overexpression mimicked shHes1-induced MET, and inhibited the migration and invasion of NPC cells, whereas shRNA-PTEN expression in shHes1-expressing CNE2 cells reversed shHes1-induced MET and abrogated shHes1mediated suppression of motility and invasion of NPC cells. In addition, it has been shown that Notch1 regulated leukemic cell growth through multiple overlapping mechanisms including downregulation of PTEN by Hes1 [77]. Furthermore, the Notch signaling indirectly regulates the PI3K pathway through a Hes1-PTEN axis during normal development of T cells [46]. These studies suggest that Hes1 is involved in different physiological and pathological processes by targeting PTEN. Taken together, Hes1 triggered EMT-like phenotypes and promoted the migration and invasion of NPC cells by downregulating PTEN, a downstream target of Hes1. However, additional Hes1 targets may also be involved in EMT, and invasion and metastasis of NPC cells.
In conclusion, we demonstrated, for the first time, that Hes1 played an important role in the pathogenesis of NPC by inducing EMT-like cellular marker alterations to promote the invasion and metastasis of NPC cells. We also provided evidence showing that Hes1 downregulated PTEN to activate the PI3K/Akt pathway, which may be one of the major mechanisms of Hes1-induced EMT-like phenotypes of NPC cells. Therefore, Hes1 may be a promising therapeutic target for the treatment of advanced NPC by inhibiting EMT, invasion and metastasis of cancer cells.

Patients and tissue samples
A total of 103 paraffinembedded NPC biopsies and 29 non-cancerous nasopharyngeal epithelial biopsies were collected in the Department of Pathology, the People's Hospital of Gaozhou City, the Affiliated Hospital of Guangdong Medical College, China, between 2003 and 2005. None of the 103 NPC patients received preoperative radiotherapy or chemotherapy. Informed consent was approved by the local Institutional Research Ethics Committee. The clinicopathologic variables and related information of NPC biopsies were collected as previously described [53,54,78].

RNA isolation and quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from NPC cells using Trizol Reagent (TaKaRa, Dalian, China) according to the manufacturer's instruction. Then mRNA was reversely transcribed to cDNA using the PrimeScript RT reagent Kit (TaKaRa). To evaluate the mRNA levels of a number of genes, qRT-PCR was performed on a Stratagene Mx3005P qRT-PCR System using SYBR Green qRT-PCR master mix (TaKaRa). GAPDH was used as the internal control. The primers used in qRT-PCR assay were listed in Supplementary Table S5. All samples were normalized to internal controls and fold changes were calculated based on relative quantification (2 −≦≦Ct ).

Transwell migration and boyden invasion assays
For transwell migration assay, 1 × 10 5 cells were seeded into the upper chamber (with 8.0 μm pores, BD Biosciences) in serum-free RPMI 1640. Boyden invasion assay was performed using matrigel (BD Biosciences) in the upper chamber. RPMI 1640 with 10% FBS was loaded in the lower compartment as chemo-attractant. After 20 hours, the migrated or invaded cells were fixed with 100% methanol, stained with hematoxylin solution (Sigma), and counted in five randomly selected optical fields.

In vivo metastasis analysis in nude mice
Female BALB/c nude mice (4-5 weeks) were purchased from the Medical Laboratory Animal Center of Guangdong Province, and maintained in microisolator cages under aseptic conditions. For in vivo metastasis assays, the vector-or Hes1-expressing CNE2 cells (1.0 × 10 6 ) were subcapsularly transplanted into the liver of nude mice (8 mice/per group), respectively. All animals were sacrificed on the thirtieth day after transplantation and liver and lung were collected for metastasis analysis. This study was carried out in accordance with the Guide for the Care and Use of Laboratory Animals of the Southern Medical University. The protocol was approved by the Committee on the Ethics of Animal Experiments of the Southern Medical University. Subcapsular transplantion of NPC cells was performed under sodium pentobarbital anesthesia to minimize suffering.

Chromatin immunoprecipitation (ChIP)
ChIP assay was performed to identify Hes1 binding sites on PTEN promoter in CNE2 cells using the Pierce Agarose Thermo ChIP Kit (Thermo) according to the manufacturer's instruction. Briefly, crosslinking was performed by adding formaldehyde (final concentration 1%) and incubated at room temperature for 10 minutes. Cross-linking reaction was terminated by the addition of glycine solution. Cells were washed with ice-cold PBS containing 0.1 mM PMSF. Cell pellets were collected by centrifugation at 3000 g for 5 minute and resuspended in 1 ml of ChIP sonication buffer. DNA was sheared by sonication and the cell debris was pelleted by centrifugation at 9,000 g for 3 minutes. Equal aliquots of chromatin supernatants were subjected to overnight immunoprecipitation with anti-Hes1 antibody (Abcam) or IgG antibody (negative control). The primer sets used in quantitative real-time PCR were listed in Supplementary  Table S6.

Statistical analyses
Statistical analyses were performed using the SPSS 13.0 software package. The χ 2 test was used to analyze the association between clinicopathological characteristics and Hes1 expression. Two-tailed Student's t test was used for comparisons of two independent groups. The data were presented as mean ± SEM. The "*" sign denotes P < 0.05 compared with control and the "**" sign denotes P < 0.01 compared with control.