Targeting STAT3/miR-21 axis inhibits epithelial-mesenchymal transition via regulating CDK5 in head and neck squamous cell carcinoma

A total of 60 HNSCC tumor samples were included in this study. All the patients received radical resection of tumor and neck lymph node dissection at Tianjin Medical University Cancer Institute & Hospital between January 2008 and December 2012. Ten tissues samples adjacent to tumors were also collected and used as controls. Institutional Human Subject Research Review Committee of Tianjin Medical University approved all procedures. All cases were grouped into positive or negative lymph node metastasis according to the American Joint of Cancer Committee’s Lip and Oral Cancer TNM staging system (2010 version) [22]. All tumor samples and tissues adjacent to tumor were paraffin-embedded and sectioned for routine immunohistochemistry (IHC) staining and in situ hybridization. Hybridization was performed with locked nucleic acid (LNA)-modified antisense miR-21 probe (Boster, China) and IHC for CDK5 (Cell Signaling Technology, USA), N-cadherin (Cell Signaling Technology, USA), Vimentin (BD Bio, CA, USA), β-catenin (Cell Signaling Technology, USA), STAT3/-p (Cell Signal Technology, USA), Ki-67 (Zhongshan, China), cleaved caspase 3 (Cell Signaling Technology, USA) and E-cadherin (Cell Signaling Technology, USA). The In situ hybridization (ISH) and IHC assays were performed as previously described [12].

The human Hep-2 and Tca8113 cell lines were purchased from the China Center for Type Culture Collection (Wuhan, China) and maintained in modified Eagle’s medium (MEM) and RPMI-1640 (Invitrogen, USA) supplemented with 10 % fetal bovine serum (Gibco, USA). WP1066 (Calbiochem, Germany) was dissolved in DMSO (Solarbo, China) for use and storage. Hep-2 and Tca8113 cells were treated with WP1066 for 48 h at a concentration of 4.5 μM. The asON was transfected by using Lipofectamine 2000 (Invitrogen, USA) at a final concentration of 300 nmol/L and 233 nmol/L, respectively, according to the manufacturer’s instructions. 48 h after transfection, cells were harvested for total RNA or miRNAs isolation, and protein extraction. The CDK5 plasmid (CDK5(BC005115)-GV141) was synthesized by Gene Chem Company, China.

Total RNAs were extracted by using the Trizol Reagent (Life technology, USA) in accordance with the manufacturer’s instructions. Both RT and PCR primers were purchased from Gene Pharma (Shanghai, China). Expression of mature miR-21 was measured using miR-21 QPCR kit (Beyotime Biotechnology, China). Relative quantification was conducted using amplification efficiencies derived from cDNA standard curves. Relative gene expressions were obtained. Quantitative analysis of change in expression levels was performed using a real-time PCR machine (7500 ABI; USA). The PCR procedure was performed on the Real-time PCR (Bio-Rad, USA) as described previously [12]. U6 was used as an internal control.

WP1066 and asON treated Hep-2 and Tca8113 cells were collected for total protein extraction. Extracted proteins were transferred to PVDF membranes (Millipore, USA) and probed with the primary antibodies against N-cadherin (Cell Signaling Technology, USA), vimentin (BD Bio, CA, USA), β-catenin (Cell Signaling Technology, USA), p-β-catenin (Cell Signaling Technology, USA), E-cadherin (Cell Signaling Technology, USA), CDK5 (Cell Signaling Technology, USA), STAT3/-p (Cell Signal Technology, USA). β-actin (Zhongshan, China) and Histone H2A (Cell Signaling Technology, USA) was used as a loading control in total cell extracts and nuclear extracts.

For immunofluorescence staining, Hep-2 and Tca8113 cells were seeded on coverslips and fixed with 4 % paraformaldehyde (PFA, Sigma), treated with 1 % Bovine serum albumin (Equitech-Bio, USA) for 30mins and incubated with the antibodies described above overnight at 4 °C. Alexa Fluor 546Goat Anti-Rabbit IgG (1:400 dilutions, Zhongshan, China) was added for 1 h at 37 °C. DAPI reagent was used to stain the cell nuclei. The cells were visualized using FV-1000 laser scanning confocal microscopes.

Hep-2, Tca8113 and Tca8113P160 cells layers were scratched using a 20 μl sterile pipette tip to form wound gaps. The wound location in the six-well plates was marked. Cells were photographed to record the wound width (0 h). Cells were cultured in serum-free medium. Photographs were taken after 24, 48 and 72-h at the marked wound location to measure the cell migration ability. Cell invasion assays were performed using transwell membranes coated with Matrigel (BD Bio, CA, USA). The lower chamber was filled with 10 % FBS. After 48 h, cells remaining in the upper chamber were removed with cotton swabs, while invading cells were fixed with 3 % paraformaldehyde (Santa Cruz, USA), stained with crystal violet (Solarbo, China). Cells penetrated through the polyethylene terephthalate membrane were counted in ten representative microscopic fields (200 × magnification).

Tianjin Medical University Animal Care and Use Committee approved all animal experimental protocols. Four nude mice were injected with 10⁷ of Tca8113 cells subcutaneously. Mice were monitored daily and three out of four mice formed tumors. Ten days after xenografting, when the tumor size reached approximately 7 mm in diameter, the tumors were surgically removed, cut into pieces of 1–2 mm³ and re-implanted into the left inguinal region of 20 recipient mice. The mice were randomly divided into four groups (DMSO group, nsON-treated group, WP1066-treated group and asON-treated group) for perspective treatment and observation. At the end of the 23^rd day of observation, the mice were sacrificed and the xenograft tumors were removed for formalin fixation and preparation of paraffin-embedded sections.

Data were analyzed using software SPSS 17.0. The interrelationship of miR-21, CDK5 and EMT related protein expression with lymph node metastasis status was analyzed using χ ² or Fisher exact test. Spearman’s correlation test was used to analyze the correlation of miR-21, CDK5 and the EMT related proteins with lymph node metastasis status. The results were expressed as mean +/- standard error (SE), which represented the average of at least three experiments and each of them was performed in triplicate. Statistics was determined using the analysis of variance test. A p value less than 0.05 was considered statistically significant.

We confirmed that all the protocols on human tissue examination and animal experiments were approved by the Committee of Medical Ethics at Tianjin Medical University.

In the present study, we determined the expression level of miR-21, CDK5 and EMT-related proteins in 60 HNSCC samples. The clinical staging was determined according to American Joint of Cancer Committee (AJCC) Lip and Oral Cancer TNM staging system (2010 version). As shown in the representative IHC staining and in situ hybridization staining in Fig. 1a and b, expression of miR-21 (red) and CDK5 was elevated in lymph node positive group compared with the negative group (χ ² = 8.906, p =0.003; χ ² = 14.102, p =0.000). MiR-21 and CDK5 were significantly associated with lymph node metastatic status (R = 0.385, p =0.002; R = 0.527,p =0.000). As shown in Fig. 1c and Additional file 1: Figure S2B, the expressions of STAT3 and EMT markers were enhanced in lymph node positive group with significant correlation to lymph node metastatic status in HNSCC samples, for N-cadherin (X ² = 21.365,p =0.000;R = 0.647, p =0.000), vimentin (X ² = 7.358, p =0.007;R = 0.350, p =0.006), E-cadherin (X ² = 21.609, p =0.000;R = -0.645, p =0.000), β-catenin (X ² = 5.742, p =0.017;R = 0.354, p =0.002). All HNSCC samples were analyzed using Kaplan-Meier analysis and a multivariate COX regression model. Analysis showed that the expression of miR-21, CDK5, EMT markers, and lymph node metastasis are closely correlated (Additional file 2: Table S1). Univariate analysis showed significant relationships between the Overall survival (OS) and higher T stage, high expression of CDK5, MiR-21, N-cadherin, Vimentin, low expression of E-cadherin, and lymph node metastasis (p < 0.05). Furthermore, T stage (HR: 1.248-3.392; p < 0.05), MiR-21expression (HR: 1.448-13.607; p < 0.05), and N-cadherin expression (HR: 1.795-38.098; p < 0.05) were identified as the independent factors of the OS, which were analyzed by Multivariate Analysis. Protein level of CDK5 and EMT markers were verified by Western blot in the aforementioned corresponding samples of HNSCC (Additional file 3: Figure S1B).

To investigate the involvement of miR-21 and CDK5 in HNSCC, we examined the expression of miR-21 in different HNSCCs cell lines (Cal-27, Hep-2, Tscca, Tca8113, and Tca8113p160). By qRT-PCR and Western blot analysis, we showed that the expression levels of miR-21 and CDK5 were remarkably higher in Hep-2 and Tca8113 cells than other HNSCCs cell lines, as showed in Fig. 2a and b. We also validated the protein level of STAT3 and pSTAT3 in all the tested cell lines, showing significantly increased pSTAT3 in Hep-2 and Tca8113 cells (Additional file 3: Figure S1A), suggesting a positive correlation of expression of miR-21, CDK5, STAT3 and activated pSTAT3. Therefore, Hep-2 and Tca8113 cells were chosen for further investigation.

To knockdown miR-21 in Hep-2 and Tca8113 cells, we used WP1066, a specific STAT3 small-molecule inhibitor, and antisense oligonucleotide (asON) to suppress miR-21 activity. MiR-21 relative expression was examined after asON exposure for optimal time and dose, the nsON was used as a control group. The optimal time and dose of WP1066 treatment in Hep-2 and Tca8113 cells were demonstrated in our previous report [16, 17]. As shown in Fig. 2c, miR-21 expression was significantly decreased after WP1066 or asON treatment. Meanwhile, CDK5 expression was suppressed by WP1066 in both Hep-2 and Tca8113 cells (Fig. 2d). Consistent with our previous report, STAT3 and pSTAT3 expression were effectively inhibited by WP1066 when compared with parental and DMSO-treated Hep-2 and Tca8113 cells (Fig. 2e).

We used asON and WP1066 to antagonize miR-21 in Hep-2 and Tca8113 cells. nsON and DMSO were used as controls. After 48 h of treatment, scratch and transwell chamber assay were employed to measure antagonistic effect in in vitro cell migration and invasion. The number of migrating and invading cells in asON and WP1066-treated cultures were reduced relative to the control cells (p < 0.05, Fig. 3a, b), demonstrating that both asON and WP1066 could impair the ability of Hep-2 and Tca8113 cells migration and invasion. Moreover, we examined the MMP-2/MMP-9 expression levels in Hep-2 and Tca8113 cells after WP1066 and asON exposure by Western blot. As shown in Fig. 3c, both MMP-2 and MMP-9 expression was decreased in WP1066 and asON groups. All these results demonstrated that knockdown of miR-21 had a suppressive effect on the migration and invasion ability of HNSCC cells in vitro. We then focused on the effect of knocking-down miR-21 on the EMT-phenotypes of Hep-2 and Tca8113 cells. Western blot showed that in both Hep-2 and Tca8113 cells treated with asON and WP1066, E-cadherin expression level was elevated while the N-cadherin expression was attenuated. In addition, a decreased expression of vimentin was also observed. Importantly, the β-catenin expression in nucleus and cytoplasm were both decreased (Fig. 4a, c), suggesting that antagonism of miR-21 by asON and WP1066 led to EMT reversal in Hep-2 and Tca8113 cells. Immunofluorescence (IF) staining showed a notable reduction of nuclear and cytoplasmic β-catenin in asON and WP1066 groups compared with the control group. We also observed inhibition of N-cadherin and vimentin in asON and WP1066-treated cells, compared with untreated cells. Meanwhile, E-cadherin exhibited opposite expression after treatment (Fig. 4d, Additional file 1: Figure S2A). The alteration of these EMT-related proteins after asON or WP1066 exposure suggests that the EMT in Hep-2 and Tscca8113 cells was suppressed.

To further elucidate that inhibition of miR-21 reversed EMT through CDK5/p35, CDK5 and p35 expression were detected by Western blot analysis. In Fig. 4b, it showed that CDK5 and p35 were significantly down regulated by asON and WP1066 in Hep-2 and Tca8113 cells, suggesting miR-21 regulates EMT process in a CDK5/p35 dependent manner.

In vitro experiments demonstrated that miR-21 reversed EMT process via CDK5/p35. We thus employed a Tca8113 xenograft tumor model to confirm this effect in vivo. The growth rates in WP1066 and asON group were suppressed compared to the control and DMSO groups (Fig. 5a). The impaired tumor growth is a consequence of both decreased proliferation and increased apoptosis as demonstrated by IHC staining of the tumors for Ki-67 and cleaved caspase 3, respectively (Additional file 1: Figure S2C). ISH analysis showed that miR-21 was inhibited by WP1066 and asON (Fig. 5d). IHC staining and western blot demonstrated decreased expression levels of STAT3, pSTAT3, CDK5, N-cadherin, vimentin and β-catenin and increased expression of E-cadherin, as observed in both WP1066- and asON-treated groups in xenograft tumors (Fig. 5b, c, Additional file 1: Figure S2C, Additional file 3: Figure S1D). We further validated that CDK5 is a required component for miR-21 to regulate EMT and invasion. CDK5 was overexpressed in Tca8113p160 cells, and we observed increased expression of EMT markers (Additional file 3: Figure S1E), enhanced migration and invasion in Tca8113p160 cells (Additional file 3: Figure S1 F, G). Moreover, CDK5 overexpression reverted the inhibition of EMT markers expression by asON and WP1066 in Hep-2 and Tca8113 cells (Additional file 3: Figure S1C), suggesting that CDK5 overexpression in HNSCC cells is sufficient to induce tumor cell motility via modulating EMT. These results not only support co-regulation of miR-21 and CDK5, but also establish a causal link connecting miR-21, CDK5 and cancer cell migration, invasion and metastasis.