CXCR7 promotes migration and invasion in head and neck squamous cell carcinoma by upregulating TGF-β1/Smad2/3 signaling

The chemokine receptor CXCR7 has been suggested to play important roles in the progression of several types of cancers. However, few studies have investigated the biological roles of CXCR7 in head and neck squamous cell carcinoma (HNSCC). CXCR7 expression and its clinical implications were examined in 103 HNSCC tissues using immunohistochemistry (IHC). The biological roles and mechanisms of CXCR7-mediated signaling pathways were investigated in HNSCC cells through CXCR7 overexpression in vitro and in vivo. High expression of CXCR7 was significantly associated with tumor size (P = 0.007), lymph node metastasis (P = 0.004), and stage (P = 0.020) in HNSCC. Overexpression of CXCR7 in HNSCC cells enhanced cell migration and invasion in vitro and promoted lymph node metastasis in vivo. CXCR7 also induced epithelial–mesenchymal transition through PI3K/AKT. CXCR7 increased secretion of transforming growth factor-β1 (TGF-β1) and promoted EMT through phosphorylated Smad2/3. Taken together, our results provide functional and mechanistic roles of CXCR7 as a master regulator of oncogenic TGF-β1/Smad2/3 signaling in HNSCC, suggesting that CXCR7 might be a therapeutic target for the treatment of HNSCC.


Results
Expression of CXCR7 and its association with clinicopathological features. To analyze the function of CXCR7 in HNSCC progression, we first evaluated CXCR7 expression by immunohistochemical analysis of tumor specimens from 103 patients with HNSCC. CXCR7 was located in the membrane and/or cytoplasm, and the intensity of the immunohistochemical staining varied as follows: negative staining (score 0), 17 cases; weak staining (score 1), 34 cases; moderate staining (score 2), 38 cases; strong staining (score 3), 14 cases (Fig. 1A). Of the 103 tumors examined, 51 (49.5%) were classified as CXCR7-low and 52 (50.5%) were classified as CXCR7high tumors. We next analyzed the correlation between CXCR7 expression and various clinicopathological factors that can affect the prognosis of patients with HNSCC. The results are summarized in Table 1. The CXCR7high group had increased tumor size (P = 0.007), lymph node metastasis (P = 0.004), and advanced tumor stage (P = 0.020). These findings indicate that high CXCR7 expression is strongly associated with advanced disease in HNSCC.

CXCR7 promotes cell migration and invasion in vitro and in vivo. To investigate whether CXCR7
promotes tumor progression of HNSCC, we created stable HNSCC cells with overexpression of CXCR7. The relative expression of CXCR7 in two different HNSCC cell lines, SNU1041 and Hep-2, was confirmed by flow cytometry, Western blotting, and immunofluorescence staining (Fig. 1B-D). Overexpression of CXCR7 significantly increased motility, migration, and invasion of HNSCC cells (Fig. 1E-G). To explore further the effect of CXCR7 overexpression on tumor metastasis in HNSCC in vivo, HNSCC cells overexpressing CXCR7 were implanted into the tongue of BALB/c nude mice. Eight weeks after cell inoculation, cervical lymph node metastasis was significantly increased in mouse xenografts with CXCR7 overexpression compared with controls (Fig. 1H). These results suggest that CXCR7 promotes LN metastasis of HNSCC cells through increased cell motility and invasion.
CXCR7 induces EMT through the PI3K/AKT signaling pathway. CXCR7-overexpressed cells displayed morphologic changes from their normal round-shaped, epithelial phenotype to a spindle-shaped, mesenchymal phenotype ( Supplementary Fig. S1A). Moreover, the expression of epithelial markers such as E-cadherin and Ep-CAM was markedly downregulated in CXCR7-overexpressed cells, whereas the expression of mesenchymal markers including N-cadherin, α-SMA, Slug, Twist, and Vimentin was upregulated ( Fig. 2A). Immunofluorescence results showed that E-cadherin was decreased and Vimentin was increased in CXCR7-overexpressed cells ( Supplementary Fig. S1B) To explore the signaling mechanisms involved in the EMT of CXCR7-overexpressed cells, we examined the AKT, ERK1/2, JNK, and p38 signaling pathways. Phosphorylation of AKT was increased in CXCR7-overexpressed cells compared to mock cells (Fig. 2B). To determine whether PI3K/AKT pathway activity was required for CXCR7-mediated EMT, mock and CXCR7-overexpressed cells were treated with the PI3K/AKT inhibitor LY294002 or wortmannin. There was no significant change in EMT markers when PI3K/AKT inhibitors were treated in mock cells (Fig. 2C). However, N-cadherin, α-SMA, Slug, Twist, and Vimentin expressions were decreased by PI3K/AKT inhibitors in CXCR7-overexpressed cells (Fig. 2C). The above results were also supported by immunofluorescence results (Supplementary Fig. S1C). Inhibition of PI3K/AKT suppressed the enhancement of cell migration and invasion, which was induced by CXCR7 overexpression (Fig. 2D,E). Taken together, these findings indicate that PI3K/AKT played a crucial role in the CXCR7-induced EMT of HNSCC cells.
CXCR7 promotes TGF-β1 signaling and Smad2/3 phosphorylation-enhanced EMT. TGF-β1, which is produced by tumor cells and immune cells, is a pleiotropic cytokine that regulates tumor progression and EMT, and is abundantly expressed in tumors 15 . In response to TGF-β, Smad2 and Smad3, which regulate target genes through interactions with other transcription factors, are activated 16 . We investigated TGF-β1-induced EMT in CXCR7-overexpressed cells. TGF-β1 is highly upregulated in CXCR7-overexpressed cells (Fig. 3A) and increased levels of TGF-β1 were detected in conditioned medium of CXCR7-overexpressed cells (Fig. 3B). In the tissue of CXCR7-high patients, there was significantly high expression of TGF-β1 (Fig. 3C). A pan-cancer analysis from The Cancer Genome Atlas (TCGA) also showed increased expression of CXCR7 and TGF-β1 in HNSCC. Moreover, there was a correlation between CXCR7 and TGF-β1 expression (Fig. 3D). We next examined whether CXCR7 overexpression affects the expression of Smad2/3, which are downstream targets of TGF-β1 signaling. The expression and phosphorylation of Smad2/3 were increased in CXCR7-overexpressed cells (Fig. 3E). In addition, treatment with recombinant TGF-β1 promoted Smad2/3 phosphorylation. Moreover, the expression of MMP2 and MMP9 was also increased in CXCR7-overexpressed cells compared to mock cells. This is known to cleave latent TGF-β1 to form active TGF-β1 (Fig. 3E,F). To confirm whether CXCR7 overexpression induces the migration and invasion of HNSCC cells in a TGF-β1 ligand dependent manner, we treated cancer cells with an anti-TGF-β1 monoclonal antibody. Neutralization of TGF-β1 suppressed the cell migration and invasion induced by CXCR7 overexpression (Supplementary Fig. S2). Therefore, the concordant expression of the CXCR7/TGF-β1 axis might highlight the importance of CXCR7 in determining the outcome of patients with cancer.
To determine whether CXCR7 overexpression induces EMT in HNSCC cells by TGF-β1/Smad2/3 signaling, we knocked down Smad2/3 expression using siRNA in CXCR7-overexpressed cells. Knockdown of Smad2/3 increased the expression of E-cadherin and Ep-CAM in Hep-2-CXCR7 cells, whereas it decreased the expression of N-cadherin, Slug, Twist, and Vimentin, as well as PI3K and AKT phosphorylation in SNU-CXCR7 and Hep-2-CXCR7 cells (Fig. 3G). These data demonstrate that TGF-β1/Smad2/3 signaling plays a crucial role in the EMT of CXCR7-overexpressed cells. The above results were also supported by immunofluorescence results (Fig. 3H).

Discussion
In this study, we demonstrated that CXCR7 is involved in TGF-β1-mediated EMT in HNSCC, resulting in tumor progression (Fig. 4F). Given the demonstrated importance of CXCR7 in tumorigenesis, we examined the expression of CXCR7 in human HNSCC tissues. CXCR7 was differentially expressed in HNSCC cells, and high expression of CXCR7 was associated with aggressive behavior of HNSCC. Overexpression of CXCR7 promoted HNSCC cell metastasis, both in vitro and in vivo. More importantly, we showed that CXCR7 regulated HNSCC cells by increasing autocrine TGF-β1 signaling, suggesting that CXCR7 could be a potential therapeutic target and predictive indicator in HNSCC.
HNSCC is characterized by high proliferation and regional lymph node metastasis with poor clinical outcome 17,18 . Lymph node metastasis is a well-known prognostic factor in HNSCC 19 . EMT transforms epithelial cells into mesenchymal cells, enabling the migration of cancer cells essential for metastasis 20 . Lymph node metastasis increases the expression of EMT-related proteins, especially in the rapidly growing margin of the lymph node 21 . HNSCC cells that have undergone EMT are more aggressive. Furthermore, loss of E-cadherin associated with EMT is an indicator of poor clinical outcome in HNSCC 22 . In our study, CXCR7 induced an EMT-like phenotype and expression of EMT-related proteins in HNSCC, and it was significantly correlated with lymph node metastasis, large tumor size, and an advanced tumor stage.
The mechanisms involved in the correlation between CXCR7 and poor prognosis have been examined in various cancers. Overexpression of CXCR7 results in increased migration and invasion [23][24][25] . Stromal cell-derived factor-1α (SDF-1α or CXCL12) is a CXC chemokine that acts as a ligand for two types of receptors, CXCR4 and CXCR7 26 . A recent study showed that SDF-1α binds to CXCR7 10-fold more than it does to CXCR4, and is involved in cell survival, adhesion, and tumor development 27 . Interestingly, CXCR7 overexpression alone increased migration and invasion ability, independently of SDF-1α 28  www.nature.com/scientificreports www.nature.com/scientificreports/ promoted cell motility, migration, and invasion in vitro, and induced lymph node metastasis in vivo, independently of SDF-1α. SDF-1α treatment also increased cell migration and invasion (data not shown). Several lines of evidence suggest that PI3K/AKT activity is important for migration and invasion in HNSCC 29,30 . Similarly, we found that CXCR7 overexpression increased PI3K/AKT phosphorylation and promoted the migration and invasion of HNSCC cells. In addition, inhibition of PI3K/AKT suppressed the enhancement of cell migration and the ability to invade, which was induced by CXCR7 overexpression. Moreover, suppression of PI3K/AKT with a PI3K inhibitor resulted in the reversal of EMT marker expression.
TGF-β acts as an early tumor suppressor in tumorigenesis. However, TGF-β becomes a major inducer of EMT during cancer progression in late-stage tumors. It even leads to tumor progression and metastasis by increasing the production of TGF-β in either an autocrine or paracrine manner 31 . In various cancers, TGF-β signaling plays www.nature.com/scientificreports www.nature.com/scientificreports/ a necessary role in the extracellular microenvironment and cellular mechanisms that promote invasion, migration, proliferation, differentiation, and apoptosis 32 . TGF-β signaling is activated through two main pathways, the canonical or non-canonical. An inactive TGF-β cytokine, which is a latent TGF-β, is located in the extracellular matrix (ECM) 33,34 . In the canonical manner, TGF-β signals via TGF-βRI phosphorylate the cytoplasmic Smad2/3 proteins, which then form a complex with Smad4 35 . This complex regulates transcription of EMT genes 36 . Moreover, TGF-β1 upregulates CXCR7 expression in a Smad2/3-dependent manner in cancer 5,37 . Although some studies have examined the upregulation of CXCR7 by TGF-β1, there has been no research showing that the Hep-2  www.nature.com/scientificreports www.nature.com/scientificreports/ expression of CXCR7 increases the expression and secretion of TGF-β1. Our study showed that TGF-β1 expression and secretion were induced by CXCR7 overexpression. Notably, CXCR7 overexpression was significantly associated with induction of Smad2/3, PI3K/AKT, MMP-2, and MMP-9 through TGF-β1 activation.
Malignant epithelial cells and stromal cells in the tumor microenvironment produce MMPs and zinc-dependent proteases. Degradation of the ECM is essential for migration and invasion of cancers. Thus, expression of proteases such as MMPs is important for HNSCC invasion 38 . In addition to the pro-tumorigenic function, MMPs have been proposed to be crucial components in the stromal activation of latent TGF-β 39,40 . We wanted to understand how CXCR7 induces TGF-β1 production, leading to EMT. Our data show that CXCR7 overexpression induced activation of MMP2 and MMP9. This finding suggests that CXCR7-mediated activation of MMP2 and MMP9 induces the activation of latent TGF-β1 and ECM degradation.
TGF-β-induced EMT has been known to transmit through both Smad-and non-Smad-dependent signals, involving crosstalk between PI3K/AKT and Smad proteins 41,42 . According to our findings, CXCR7 overexpression enhanced TGF-β1-induced cell migration, invasion, and EMT. CXCR7 induced the activation of PI3K/AKT but not of MAPKs including ERK, JNK, and p38 MAPK. Furthermore, inhibition of TGF-β signaling with siRNA for Smad2/3 suppressed activation of PI3K/AKT and restored EMT marker expression. However, suppression of PI3K/AKT with a PI3K inhibitor did not influence Smad2/3 phosphorylation (data not shown). These results imply that CXCR7 partially activates PI3K/AKT through TGF-β mediated Smad2/3 activation. Thus, we demonstrated that CXCR7 is a master regulator of oncogenic TGF-β1/Smad2/3 signaling in HNSCC.
In conclusion, we demonstrated that CXCR7 plays a key role in the progression of HNSCC. Overexpression of CXCR7 in HNSCC was significantly associated with tumor metastasis. Furthermore, functional and mechanistic studies revealed that CXCR7 regulated EMT and ECM remodeling to achieve higher motility and invasiveness through the activation of TGF-β1/Smad2/3 signaling. Therefore, our study suggests that CXCR7 may act as an effective prognostic marker and a promising treatment target in HNSCC.  Specimen preparation and immunohistochemistry. Specimen preparation and immunohistochemistry was conducted as previously described 43 . For immunohistochemistry, 3-μm-thick sections were cut from the recipient blocks. All procedures were performed at room temperature, as recommended by the manufacturer. In brief, sections were dewaxed in xylene and then rehydrated in graded alcohols. Sections were washed in water before antigen retrieval using a Dako PTLink machine (Dako, Glostrup, Denmark) with 10 mM sodium citrate buffer (pH 6.0) at 97 °C for 20 min. Sections were treated with 3% hydrogen peroxide for 10 min to block endogenous peroxidase and preincubated with a serum-free protein block solution (Dako) for 20 min to eliminate background staining. Prepared polyclonal mouse antibodies raised against human CXCR7 (Abcam, Cambridge, UK; ab38089) and TGF-β1 (ABcam; ab169771) were diluted at 1:1000 with background-reducing diluents (Dako). Samples were incubated overnight at 4 °C in a humidified chamber and washed with TBS-T. Slides were then incubated for 30 min with an EnVision anti-mouse (Dako) polymer. Reaction products were visualized with diaminobenzidine (DAB) plus substrate-chromogen solution for 5 min. Slides were counterstained with Meyer's hematoxylin and mounted. Careful rinses with several changes of phosphate buffered saline (PBS) were performed between stages of the procedure. Negative controls consisted of excluding the primary antibody or using preimmune IgG1 to evaluate nonspecific staining.
Evaluation of immunohistochemical staining. Evaluation of immunohistochemical staining was conducted as previously described 43 . Immunohistochemical staining results were evaluated by two independent pathologists who were blinded to the patients' clinicopathological details. The immunohistochemical staining was categorized according to the following scoring method, which included four grades based on staining intensity: 0, no staining; 1, weak staining; 2, moderate staining; and 3, strong staining. In the case of heterogeneous staining within samples, the higher score was selected if more than 50% of cells showed higher staining intensity. For all patients, scores from two tumor cores in the same patient were averaged to obtain a mean score. Cases with staining intensity scores of 0-1 were placed in the CXCR7 low-expression group, and those with staining intensity scores of 2-3 were placed in the CXCR7 high-expression group. CXCR7 overexpression in HNSCC cell lines. CXCR7 overexpression in HNSCC cell lines was conducted as previously described 44 . Overexpression of CXCR7 in HNSCC cells was achieved through the use of lentivirus-mediated transduction of full-length human CXCR7 subcloned into a pLVX-EF1 α-IRES-Puro lentiviral vector (Clontech, Mountain View, CA, USA). To generate a stable transfectant, the acquired lentiviral vector was co-transfected into 293 T cells with virus packaging mix (Sigma Aldrich) using a Lipofectamine 3000 (Invitrogen) according to the manufacturer's protocol. The virus was harvested from the supernatant and concentrated with lenti-X-concentrator (Clontech), then added to SNU1041 and Hep-2 cells along with 5-µg/mL polybrene (Santa Cruz). Puromycin-resistant cells were selected by culture for 2 weeks in the presence of puromycin. CXCR7 expression levels were analyzed by flow cytometry (Beckman Coulter, Brea, CA, USA) and Western blot analysis.
Migration and invasion assays. Migration and invasion of HNSCC cells were implemented using a 8-µm pore size transwell chamber (Corning Costar, Cambridge, MA). Migration and invasion assays were performed using previously reported protocols 44 . Briefly, the lower surface of the transwell was coated with 0.1% gelatin (Sigma Aldrich) for the migration assay, and the upper side was coated with 25 µg/ml of matrigel for the invasion assay (Matrigel; BD Biosciences, Franklin Lakes, NJ, USA). Fresh culture medium containing 10% FBS was placed in the lower chamber as a chemo-attractant. HNSCC cells were suspended at a final concentration of 1 × 10 5 cells/ ml in medium containing 0% FBS. Cell suspensions were loaded into each of the upper wells, and the chamber was incubated at 37 °C for 24 h (migration) or 48 h (invasion). Cells were fixed and stained with 0.1% crystal violet staining. Chemotaxis activity was quantified by counting the cells that migrated to the lower side of the filter with a microscope. Five randomly chosen fields were counted for each assay.
Western blot analysis. Western blotting was performed using our previously reported protocols 44 . Briefly, cells were lysed in ProEX TM CETi Lysis buffer (TransLab, South Korea) with protease inhibitor cocktail (Sigma Aldrich) and phosphatase inhibitor cocktail (Roche). Cell lysates were separated using SDS-PAGE and then transferred to polyvinylidene difluoride (PVDF) membranes (PALL). The membranes were incubated with the indicated primary antibodies, followed by incubation with horseradish peroxidase-conjugated secondary antibodies (Cell Signaling Technology). The immunoreactive polypeptides were detected using an ECL substrate (Bio-rad and Thermo Fisher Scientific).
Immunofluorescence. Immunofluorescence was performed using our previously reported protocols 44 .
Briefly, cells were attached to a chamber slide TM (Lab-TekII). After starvation for 6 h in serum-free medium, cells were fixed in 10% formalin for 10 min at 37 °C, permeabilized with 0.5% Triton X-100 (Sigma Aldrich) for 20 min at room temperature, washed in PBS, and then blocked in 3% chicken serum albumin (CSA) in PBS for 30 min at room temperature. Cells were incubated with anti-E-cadherin, anti-Vimentin and anti-CXCR7, respectively, overnight at 4 °C. The primary antibody was removed and washed and then treated with a fluorescence-conjugated secondary antibody for 2 h at 37 °C. Cells were stained with DAPI (Vector Laboratories, Burlingame, CA, USA) for nuclei staining, and coverslips were mounted on the slides.
Orthotopic model of head and neck cancer. All animal experiments were approved by the Animal Experimental Ethics Committee of Chungnam National University and animal care was performed in accordance with the guideline. Head and neck tumors in mice were generated by modifying orthotopic xenograft models as previously described by Kawashiri et al. and Sano et al. 45,46 . Inoculation with SNU1041-mock and SNU1041-CXCR7 cells was performed in five mice. In preparation for inoculation, cells were resuspended in RPMI1640 at a concentration of 1 × 10 7 cells/100 μl. Inoculation was performed by injecting 20 µl of cell suspension to deliver 2,000,000 cells into the submucosa of the dorsal tongue at the circumvallate line. The animals were euthanized 8 weeks after inoculation and the cervical lymph nodes were removed by microsurgical dissection. Specimens were formalin-fixed, paraffin-embedded, serially sectioned on 200 µm sections, and stained with H&E.
Enzyme-linked immunosorbent assay analysis. Conditioned medium from each group was collected and stored at −80 °C before enzyme-linked immunosorbent assay (ELISA) analysis. The active form TGF-β1 was measured using a human TGF-β1 ELISA kit (B&D Systems) according to the manufacturer's instructions.
Gelatin zymography assay. The gelatin zymography assay was previously described by HW Yeh et al. 47 .
Conditioned medium from each group was collected and concentrated using a centricon (Pall Corporation, 30 K). Samples were mixed with 2X sample buffer (TransLab, South Korea) and separated by 8% SDS-PAGE containing 0.1% gelatin (Sigma Aldrich). After electrophoresis, gels were washed two times in washing buffer (2.5% Triton-X 100 in TBS) and then incubated overnight at 37 °C in 1X zymogram development buffer (Bio-Rad; 161-0766). After incubation, the gels were washed and stained with EZBlue TM Gel Staining Reagent (Sigma Aldrich; G1041) and then washed until white bands appeared on the blue background.
Statistical analysis. The expression analysis was performed using Gene Expression Profiling Interactive Analysis (GEPIA). Based on TCGA Datasets, the GEPIA website (http://gepia.cancer-pku.cn/) was used for correlation between CXCR7 and TGF-β1. The χ 2 test and linear-by-linear association were used to assess correlations between CXCR7 expression and clinicopathological features. All analyses were conducted using the SPSS version 17.0 software program (SPSS, Chicago, IL). Data from in vitro and in vivo experiments were expressed as the mean ± standard error of the mean (SEM). Differences between groups were analyzed using the Student's t-test. A P-value < 0.05 was considered to indicate statistical significance. Data are representative of at least three independent experiments.