Extracellular Vesicles from Carcinoma-Associated Fibroblasts Promote Epithelial-Mesenchymal Transition of Salivary Adenoid Cystic Carcinoma via Interleukin-6

Carcinoma-associated broblasts (CAF) play a pivotal role in cancer progression. Salivary adenoid cystic carcinoma (SACC) has a high tendency to invade and metastasize. Understanding how CAF interact with SACC cells is essential to develop new targeting therapies for SACC. Extracellular vesicles (EVs) play important roles in intercellular communication. However, the role of CAF-derived EVs in SACC invasion remains poorly understood. In this study, we show that CAF EVs promote the migration and invasion abilities of SACC cells. The expression of biomarkers of epithelial-mesenchymal transition (EMT) was higher in SACC cells treated with CAF EVs treated with CAF EVs than in the negative controls, and high levels of IL-6 were detected in CAF and their EVs. Knockdown of IL-6 in CAF decreased invasive abilities and EMT biomarker expression in SACC cells induced by CAF EVs. CAF EV-associated IL6 promoted SACC EMT by activating JAK2/STAT3 signaling pathway. These ndings suggest that targeting CAF-derived EVs may be an effective strategy for inhibiting SACC invasion.

Extracellular vesicles (EVs) are nano-sized membranous vesicles. EVs carry biological cargo, such as proteins, microRNA, mRNA, and DNA [18]. EVs transfer their cargo to recipient cells, thereby altering the biological function of the recipient cells. CAF-derived EVs are involved in cancer progression and may serve as diagnostic and prognostic biomarker [19]. In prostate cancer, CAF EVs induced EMT of cancer cells by transferring miR-409 to cancer cells [20]. In colorectal cancer, CAF EVs contributed to the chemoresistance to 5-uorouracil and oxaliplatin [21].
In this study, we demonstrated that CAF EVs are involved in the EMT of SACC and promote tumor invasion. CAF EVs promoted SACC cell invasion and EMT by delivering interleukin (IL)-6 and activating JAK2/STAT3 signaling pathway. The results suggest that CAF-derived EVs are potential targets for stroma-oriented therapy in SACC.
EVs isolation CAF cells were cultured in 75cm 2 asks till cell con uency reached more than 80%. Then, they were washed with PBS and incubated with serum-free DMEM/F12 for 48 hours. The supernatant was collected as condition medium (CM) and stored at -80 Cº until use. CM was sequentially centrifuged at 500 g, 2,500 g and 10,000 g and the supernatant was collected. Then EVs were isolated from the supernatant using Total Exosome Isolation Reagent (Invitrogen 4478359).

Transmission electron microscopy (TEM)
EVs were evaluated for their morphology and size by TEM. Brie y, 40 µL of EV samples were placed on a lm, then copper mesh is applied on it for 15 min, after that the copper was drained with lter paper and 3% of phosphotungstic acid was put on the mesh for 5 minutes. Finally lter paper was used to drain the negative dye solution and lter the mesh. Then the sample was observed under the electron microscope JEM-2000EX9 (JEOR, JAPAN).

Wound healing assay
The migration abilities of both SACC-83 and SACC-LM cells were assessed by wound healing assay. Cells were seeded in 6-well plate till 100% con uency, then a straight scratch was made by using P1000 pipette tip. Then cells were washed with PBS and cultured with CAF EVs (10 µg/well) in serum free DMEM/F12 for another 48 hours. DMEM/F12 without EVs was used as a negative control. Images were recorded by an inverted uorescence microscope (Olympus IX71). The gap closure was evaluated as the cell migration rate. The relative closure rate of each sample was measured using Image-Pro Plus 6.0.

Transwell invasion assay
We assessed the invasion ability of both SACC-83 and SACC-LM cells by transwell invasion assay. It was performed using the transwell chamber consisting of 6.5 mm diameter inserts (Corning Inc., USA). The membrane with 8.0 µm-pore-size was coated with 1:10 diluted Matrigel (Corelle Life Science Co., Ltd). Brie y, SACC-83 and SACC-LM cells were seeded in the upper chamber and CAF EVs (10 µg/well) was added to the lower chamber for the induction. After 48 hours, the non-invading cells in the upper chamber were removed by cotton swaps and the cells that invaded to the bottom chamber were xed with 4% paraformaldehyde, stained with 0.1% crystal violet (coolaber science and technology), and photographed with an inverted microscope (Olympus IX 71). Finally, we took images from ve random elds and counted the cells in each chamber to get the invaded cell number as the mean. The data represent at least three experiments (mean ± standard error).

Subcutaneous xenograft model
The use of BALB/c nude mice (3-4 weeks, female) were approved by the Institute Animal Care and Use Committee of Dalian Medical University. Animal experiments, transportation, and care were conducted in compliance with the relevant laws and the guidelines issued by the Ethical Committee of Dalian Medical University. Mice were divided into three groups with ve mice in each group: SACC-LM group was injected with 2.25 × 10 6 cells/mouse; SACC-LM+CAF-A1 group was injected with 2.0 × 10 6 SACC-LM and 2.5 × 10 5 CAF-A1 per mouse; SACC-LM+CAF-A2 group was injected with 2.0 × 10 6 SACC-LM and 2.5 × 10 5 CAF-A2 per mouse. Cells were resuspended in 50µL PBS and 50µL Matrigel, then injected subcutaneously. The mice were raised for 5 weeks. Mice were anesthetized with 20% urethane when maximum tumor diameter reached to 15 mm. After euthanizing, the tumors were harvested and xed with 4% paraformaldehyde and 30% sucrose solution overnight, then embedded into OCT and prepared into sections.

Immunohistochemical staining
Immunohistochemical staining of xenografts was performed on 8-µm-thickness sections using SPlink Detection Kits (SP-9000, ZSGB-BIO, China). The sections were rinsed with PBS and incubated with 3% hydrogen peroxide in methanol for blocking endogenous peroxidase activity. The nonspeci c binding sites were blocked with 10% goat serum for 30 minutes. Sections were incubated with anti-human pan-CK

Statistical analyses
Statistical analysis was performed using SPSS 13.0 software. The one-way analysis of variance (ANOVA) is used to determine any statistically signi cant differences. Signi cance was de ned at p< 0.05.
A wound healing assay was performed to investigate the effect of CAF EVs on the migration ability of SACC cells. CAF-A1/A2 EVs markedly promoted the migration of SACC-83 and SACC-LM cells compared with that in the negative control (Fig. 1d). A transwell invasion assay demonstrated that the invasion of both SACC-83 and SACC-LM cells increased signi cantly by CAF-A1/A2 EVs treatment signi cantly increased the invasion of both SACC-83 and SACC-LM cells compared with that in the control (Fig. 1e).

CAF EVs induced EMT of SACC cells
Because EMT contributes to cancer invasion, we hypothesized that the increased migration and invasion abilities of SACC induced by CAF EVs were associated with the acquisition of an EMT state. The morphological changes and the intercellular junction between SACC cells were examined in vitro, and the expression of EMT markers was assessed by western blotting. SACC-83 and SACC-LM cells were cultured in media supplemented with CAF-A1 or CAF-A2 EVs for three days, and serum-free medium was used as the control. The results showed that SACC-83 (Fig. 2a) and SACC-LM (Fig. 2b) cells treated with CAF-A1 and CAF-A2 EVs underwent morphological changes, including the development of elongated processes resembling the morphology of mesenchymal cells. Western blot analysis of the expression of EMT biomarkers showed that the epithelial biomarker E-cadherin was signi cantly downregulated, whereas the mesenchymal biomarkers vimentin and N-cadherin were upregulated in both SACC-83 and SACC-LM cells, compared with the control group (Fig. 2c-d). Overall, these data suggest that CAF EVs promote migration and invasion of SACC cells by modulating EMT.
CAF EVs-associated IL-6 is critical in promoting SACC EMT The ndings showing that CAF EVs promoted the EMT of SACC cells prompted us to investigate the underlying mechanism. The results of cytoimmuno uorescence staining and western blotting showed high expression of IL-6 in CAF-A1 and CAF-A2 cells and their EVs (Fig. 3a). Because IL-6 plays an essential role in EMT of various cancers, we investigated whether CAF EVs-associated IL-6 promoted SACC EMT. First, CAF were transfected with a speci c siRNA against IL-6 to knock down its expression, and EVs were isolated from the CM of transfected cells. SiIL-6 transfection signi cantly downregulated IL-6 in CAF-A1 and CAF-A2 cells as well as in their EVs compared with negative control (Fig. 3b). SACC cells were then treated with EVs from transfected CAF and migration and invasion were assessed. CAF siIL-6 EVs signi cantly inhibited the migration and invasion abilities of both SACC-83 and SACC-LM cells, as determined by the wound healing assay and Transwell invasion assay, respectively (Fig. 3c-d).
Collectively, these results indicate that CAF EV-associated IL-6 plays an important role in the invasion and migration process in SACC.
Analysis of EMT biomarkers in SACC cells treated with CAF siIL-6 EVs showed that SACC-83 (Fig. 4a) and SACC-LM (Fig. 4b) cells maintained the morphological features of epithelial cells, characterized by tight intercellular junctions, compared with CAF EVs. In addition, CAF siIL-6 EVs induced an EMT phenotype characterized by increased E-cadherin expression and decreased N-cadherin and vimentin expression in SACC-83 (Fig. 4c) and SACC-LM (Fig. 4d) cells. These results suggest that IL-6 is critical for CAF EVsinduced SACC EMT.

CAF promote EMT by IL-6 in vivo
To determine whether CAF-derived IL-6 was involved with EMT of SACC in vivo, a subcutaneous xenograft model was established in BALB/c nude mice, which were divided into SACC-LM, SACC-LM+CAF-A1 and SACC-LM+CAF-A2 groups. By 12 weeks, the maximum tumor diameter reached to 15 mm. Survival analysis demonstrated that mice with SACC-LM+CAF-A1/A2 transplantation had lower survival times than those with SACC-LM transplantation solely [22]. Hematoxylin and eosin staining con rmed the cribriform growth pattern of SACC xenografts in the three groups (Fig. 6a). Immunohistochemical staining of pan-CK and vimentin identi ed tumor epithelial cells and mesenchymal cells (Fig. 6b-c). Both SACC-LM+CAF-A1 and SACC-LM+CAF-A2 groups contained more stromal components than SACC-LM group. IL-6 expression was higher in the SACC-LM+CAF-A1 and SACC-LM+CAF-A2 groups than in the SACC-LM group (Fig. 6d). IL-6 showed strong expression in tumor cells in SACC-LM+CAF-A1 and SACC-LM+CAF-A2 groups and weak expression in tumor cells in the SACC-LM group. We hypothesize that CAF-A1/A2 cells might deliver IL-6 to adjacent tumor cells via EVs.
Assessment of EMT markers showed that E-cadherin expression was lower and N-cadherin expression was higher in the SACC-LM+CAF-A1 and SACC-LM+CAF-A2 groups than in the SACC-LM group (Fig. 6e-f). These data suggest that CAF induce EMT of SACC in vivo by delivering IL-6 to tumor cells.

Discussion
SACC is one of the most aggressive salivary gland cancers and it is characterized by an invasive growth pattern and distant metastasis [23,24]. SACC invasion is related to the expression of several factors, such as EGFR, HIF-1α, and survivin [25][26][27]. The tumor microenvironment is an essential mediator of the interaction between different cell types and it is involved in the regulation of tumor invasion. CAF are critical stromal components that promote cancer cell proliferation and invasion by expressing growth factors and chemokines and depositing tumorous ECM [28]. We recently demonstrated that CAF promote SACC lung metastasis by creating a pre-metastatic niche in the lung [22]. CAF promote cancer progression by transferring factors and genes to tumor cells via EVs, such as delivering CD9 in scirrhous-type gastric cancer [29] and delivering annexin A6 in pancreatic cancer [30]. In this study, we demonstrated that CAF EVs increase the migration and invasion abilities of SACC cells by transferring IL-6 to tumor cells.
Increasing evidence indicates that EMT is associated with cancer growth, invasion, angiogenesis, and metastasis [31]. In this study, SACC cells stimulated with CAF EVs acquired EMT-like characteristics with enhanced invasion ability. EVs can induce EMT by promoting the trans-differentiation of epithelial cancer cells into a mesenchymal phenotype mediated by multiple factors [32][33][34]. IL-6 is a multifunctional cytokine that has immunological and in ammatory functions, and high expression of IL-6 is detected in different epithelial tumors [35]. IL-6 is associated with EMT induction in different cancers [36]. CAF produced IL-6 in vivo in virous cancers to induce growth and invasion through EMT [37,38]. In the present study, SACC-derived CAF secreted EVs expressing high levels of IL-6 were involved in the regulation of SACC EMT. A previous study demonstrated that IL-6 sowed high expression in parotid gland SACCs, but not in submandibular gland SACCs [39]. Hoffmann et al. and his colleagues reported that IL-6 serum levels increased in some SACC patients, compared with healthy controls [40]. Besides IL-6, other cytokines may contribute to SACC invasion and metastasis, such as transforming growth factor-β1 [41], macrophage migration inhibitory factor [42], and tumor necrosis factor receptor-associated factor 6 [43], et al. Further studies are necessary to elucidate the exact roles of cytokines on SACC invasion. IL-6 exerts its biological functions by binding to its receptor sIL-6R to activate JAK/STAT3 signaling pathway, which controls the occurrence and development of tumors [44][45][46]. In the present study, IL-6 expressed by CAF EVs activated the JAK2/STAT3 signaling pathway in SACC cells. Constitutive activation of JAK2/STAT3 is directly related to tumor progression via EMT induction. The present data indicated that the ability of IL-6 to stimulate SACC cell invasion is regulated, at least in part, by the JAK2/STAT3 pathway.