Pancreatic cancer‐derived small extracellular vesical ezrin activates fibroblasts to exacerbate cancer metastasis through STAT3 and YAP‐1 signaling pathways

Cancer‐associated fibroblasts (CAFs), a major component of the tumor microenvironment (TME) in pancreatic ductal adenocarcinoma (PDAC), play an important role in tumorigenesis, metastasis, and chemoresistance. Tumor‐derived small extracellular vesicles (sEVs), which mediate cell‐to‐cell communication between cancer cells and fibroblasts, are also critical for cancer progression and metastasis. However, it remains unclear how PDAC cell‐derived sEVs activate fibroblasts, which contributes to tumor progression. Here, we report that ezrin (EZR) expression in PDAC cell‐derived sEVs (sEV‐EZR) can activate fibroblasts, resulting in increased migration ability and high expression of α‐SMA, PDGFRB, and high production of extracellular matrix in fibroblasts. Reciprocally, sEV‐EZR‐activated fibroblasts enhanced PDAC cell proliferation, invasion, and metastasis to the liver in animal models. Conversely, fibroblasts treated with PDAC cell‐derived sEVs with EZR knockdown resulted in the reduced metastatic ability of PDAC. Mechanistically, we demonstrated that PDAC cell‐derived sEV‐EZR increases the STAT3 and YAP‐1 signaling pathways to induce fibroblast activation, and the activated fibroblasts promote PDAC cell proliferation, invasion, and liver metastasis. Inhibition of the STAT3 and YAP‐1 signaling pathways by gene knockdown can abrogate sEV‐EZR‐induced effects. These findings suggest that targeting the interaction between PDAC cell‐derived sEV‐EZR and fibroblasts is a potential therapeutic strategy for PDAC.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive human malignancies, with clinical characteristics of unnoticeable alarming symptoms for early diagnosis, advanced-stage manifestations, and resistance to chemotherapy [1]. The tumor microenvironment (TME) of PDAC is characterized by prominent desmoplastic stroma with excessive amounts of fibroblasts, collagen, laminin, and fibronectin, which construct a hypovascular microenvironment that impairs drug delivery and increases chemotherapeutic resistance, actively contributing to tumor development and metastasis [2,3]. Cancer-associated fibroblasts (CAFs) in the TME play a key role in regulating cancer cell proliferation, invasion, and metastasis [4]. Cancer-associated fibroblasts can have roles in promoting or retarding tumorigenesis depending on their independence or association with cancer cells [5,6]. Cancer-associated fibroblasts could promote cancer cell crosstalk, influence extracellular matrix deposition, and direct the immune system to create a cancerpromoting environment. By contrast, some fibroblasts inhibit cancer initiation, affect epithelial cell differentiation, and limit cancer cell invasion [7,8]. Studies have demonstrated more than one CAF population present in the TME of PDAC [9]. Activated PSCs express alpha-smooth muscle actin (a-SMA), fibroblast activation protein (FAP), platelet-derived growth factor-receptor A (PDGFRA), platelet-derived growth factor-receptor B (PDGFRB), and vimentin in the desmoplastic stroma [6,10,11]. Earlier studies also reported a positive correlation between CAF infiltration and poor outcomes in patients with cancer, including PDAC [11][12][13][14].
Extracellular vesicles (EVs) are a heterogenous group of cell-derived membrane-bound vesicles that vary in size, composition, intracellular biogenesis, and cell of origin. A single cell can deliver a large range of EVs, including small extracellular vesicles (sEV or exosomes; 30-100 nm), microparticles (200 nm-1 lm), and apoptotic bodies, which have different protein profiles due to their different routes of formation [15,16]. Small extracellular vesicles (or exosomes) released from various cells, act as mediators in intercellular communications by carrying information cargo, such as lipids, proteins, and nucleic acids, into recipient cells [17]. sEVs are enriched with a distinct population of proteins considered sEV markers, such as the transmembrane tetraspanin family (CD9, CD63, and CD81); proteins of endosomal sorting complexes, such as TSG101 and Alix; heat-shock proteins (HSP70 and HSP90); and major histocompatibility complex class I [18,19]. Previous studies have shown that sEVs can create an immunosuppressed microenvironment that promotes tumor progression via communication between tumor cells and the surrounding stromal cells [20][21][22]. In addition, sEVs can regulate the interaction between tumor cells, immune cells, and fibroblasts resulting in protumorigenic or antitumorigenic responses [23][24][25]. Accumulating evidence has also pointed out that sEVs play a significant role in the crosstalk between CAFs and cancer cells and contribute to cancer progression and metastasis [26][27][28]. Specific sEVs released from CAFs can be internalized by cancer cells, and sEVs released from cancer cells can induce the transformation of normal fibroblasts into CAFs [29]. In breast cancer, tumor-secreted sEVs could transfer miR-9 into normal fibroblasts. The uptake enhances the switch of normal fibroblasts to the CAF phenotype, thus contributing to tumor growth [26]. In melanoma, sEV-miR-155 induces the proangiogenic switch of CAFs by inhibiting SOCS1 expression and activating the JAK2/STAT3 signaling pathway [30]. A study has shown that sEVs from PDAC cells promote PSC recruitment through activating the Lin28B/let-7/HMGA2/PDGFB signaling pathway [31]. However, it is still unclear how the reciprocal effects of tumor-derived sEVs affect fibroblast activation and tumor progression in PDAC. In this study, we aimed to investigate whether PDAC cell-derived sEVs are involved in fibroblast activation and to explore the underlying mechanisms that exacerbate PDAC progression and metastasis attributed to the crosstalk between fibroblasts and PDAC cells.

Cell lines
Nontransformed human pancreatic ductal epithelial cells (HPDE) from a pancreatic specimen of a 75-yearold male [32,33] (RRID: CVCL_S973, a gift from Dr Tze-Sing Huang National Health Research Institutes, Taiwan) were grown in keratinocyte serum-free medium with 0.2 ngÁmL À1 epidermal growth factor (EGF) and 30 lgÁmL À1 bovine pituitary extract (Invitrogen Life Technologies, Waltham, MA, USA). Nontransformed human pancreatic acinar-to-ductal epithelial-like cells (HPNE) were purchased from the ATCC (Cat# CRL-4023, RRID: CVCL_C466). HPNE cells were grown in a medium containing one volume of M3 base medium (INCELL) together with three volumes of glucose-free DMEM supplemented with 5% FBS, 5.5 mM glucose, 10 ngÁmL À1 EGF, and antibiotics (penicillin/streptomycin). Pan18 cells were generated from liver metastatic sites of Elas-CreER; LSL-Kras G12D/+ ; Trp53 f/+ male mice (a gift from Dr Chia-Ning Shen, Genomic Research Center, Academia Sinica, Taipei) [34]. Human embryonic kidney 293T cells (RRID: CVCL_0063) and human pancreatic cancer cells (BxPC-3) were obtained from the ATCC (Rockville, MD, USA; Cat#CRL-1687, RRID: CVCL_0186). PDAC patient-derived PC080 and PC084 cell lines were established as previously described [35]. 293T cells and Pan18 cells were cultured in high-glucose-DMEM; BXPC-3, PC080, and PC084 cells were cultured in RPMI1640 (Gibco, Grand Island, NY, USA). These two media were supplemented with 10% exosomedepleted fetal bovine serum, penicillin and streptomycin (100 IUÁmL À1 and 100 lgÁmL À1 , respectively), 1 mM sodium pyruvate, and 1% nonessential amino acids (Gibco). Human pancreatic stellate cells (HPaSteC) were purchased from ScienCell (Catalog #3830, ScienCell Research Laboratories, Inc., Carlsbad, CA, USA) and were cultured in Stellate Cell Medium (SteCM, #5301) supplemented with 2% FBS, 1% Stellate Cell Growth Supplement (SteCGS, #5352), and 1% penicillin/streptomycin mixture (P/S, #0503; ScienCell Research Laboratories, Inc.). Primary cultures of CAFs from human PDAC specimens, including CAF041 and CAF067, were collected from surgically resected PDAC specimens between November 2011 and December 2015 according to our previous study [35], and were maintained in Stellate Cell Medium (SteCM, #5301) supplemented with 2% FBS, 1% Stellate Cell Growth Supplement (SteCGS, #5352), and 1% penicillin/streptomycin mixture (P/S, #0503) (ScienCell Research Laboratories, Inc.). All cells were cultured at 37°C in a 5% CO 2 atmosphere and maintained within 3 months of resuscitation from the frozen aliquots with less than 20 passages for each experiment. The cell lines were authenticated by morphology or by Short Tandem Repeat (STR) analysis by analyzing multiple locations within the genome containing short DNA sequence repeats. The resulting DNA profiles were used to confirm the identity and purity of all cell lines through comparison to the STR reference database in the past 3 years. In addition, all cells were regularly checked for mycoplasma infection and cell morphology to keep cells healthy. Human PDAC tissue samples were obtained from patients at the National Taiwan University Hospital (NTUH) with the understanding and written consent of each subject. The study was approved by the National Taiwan University Hospital (NTUH) Institutional Review Board (approval number 201303029RINC and 201411085RINB). The study methodologies conformed to the standards set by the Declaration of Helsinki.

Isolation and culture of mouse primary pancreatic acinar cells and fibroblasts
All mouse primary acinar cells were isolated and cultured as described by Gout et al. [36]. Briefly, 8-weekold male C57BL/6 (B6) mouse pancreases were mechanically and enzymatically digested with collagenase IA solution (19 HBSS containing 10 mM HEPES, 200 unitsÁmL À1 of collagenase IA, and 0.25 mgÁmL À1 trypsin inhibitor) to obtain isolated acinar structures. Acini were grown in a medium containing one volume of glucose-free DMEM and one volume of F12 medium supplemented with 2.5% FBS, 1% penicillin/streptomycin mixture, 0.25 mgÁmL À1 of trypsin inhibitor, and 25 ngÁmL À1 of recombinant human EGF. After being cultured on type I collagencoated six-well culture dishes for 6 days, acinar-toductal-like cells were trypsinized and replated into another type I collagen-coated plate for subsequent experiments. Primary fibroblasts from the pancreas of 6-to 8-week-old B6 mice were isolated and cultured according to a previous report [37].

Separation and characterization of sEVs from cells
The sEVs from the cell-conditioned medium of HPDE, BXPC-3, PC080, PC084, Pan18, and mouse pancreatic acinar cells were collected as previously described [38]. The size distribution and concentration of the isolated sEVs were determined as described previously [38].

Protein extraction and western blot analysis
Total proteins were extracted from cultured cells in a lysis buffer containing 50 mM Tris-HCl, 1% NP-40, 150 mM NaCl, 0.1% SDS, 1 mM PMSF, 1 mM Na 3 VO 4 , and 1 lL protease inhibitor cocktail (Sigma-Aldrich, Inc., St. Louis, MO, USA). Protein concentrations were determined by the BCA assay kit (Thermo, Carlsbad, CA, USA) with bovine serum albumin as standard. Identical amounts of protein lysates were subjected to 10-12% SDS/polyacrylamide gels and trans-  (Table S1). The membranes were exposed using enhanced chemiluminescence reagents (PerkinElmer, Waltham, MA, USA) for the HRP-coupled secondary antibodies and were analyzed using the BioSpectrum 810 Imaging System (UVP, Upland, CA, USA). Protein levels were detected as the integrated area (pixels) of the band intensities using densitometry analysis with IMAGE J software (Bethesda, MD, USA). The numerical values for protein band intensities were corrected with the values for the GAPDH bands.
Transfection was performed using TransIT-X2 Transfection Reagent (Mirus, Madison, WI, USA) according to the manufacturer's instructions. For lentivirus production, 293T cells were co-transfected with pMD.G, pCMV▵R8.91, and Lentiviral transfer plasmids using the Polyjet transfection reagent (SignaGen Lab, Ijamesville, MA, USA). We harvested the culture medium containing lentivirus 48 h after transfection and removed the cells from the culture medium by centrifugation at 500 g for 5 min and filtered it through a 0.45-lm filter. The lentiviral stocks in the medium containing polybrene (8 lgÁmL À1 ) were used to infect the target cells for 6 h.

Transwell migration and invasion assays
Transwell migration and invasion assays were performed using a Transwell insert plate (PET membrane, 8-lM pore size, SPL 36224, Korea). The transwell invasion assay was conducted after coating the insert membrane with 5 mgÁmL À1 Matrigel (Corning Matrigel Membrane Matrix, cat# 356234), whereas the transwell migration assay was conducted without coating. HPaSteC migration assay: HPaSteCs (2 9 10 4 cells per well) treated with sEV (5 lgÁmL À1 ) were seeded in the upper chamber in serum-free medium, and 600 lL of medium containing 2% FBS was added to the lower well. Cells were plated in the top chamber and incubated for 24 h to allow the cells to attach. Cancer cell invasion assay: HPaSteC (1 9 10 5 cells per well) was seeded in the lower chamber, which was filled with 600 lL of medium. PDAC cells (5 9 10 4 PDAC cells) were seeded into the upper chamber. After 24 h of incubation, migrated/invading cells were fixed on the membrane with 70% ethanol for 60 min at 4°C, washed twice with PBS, and before the interiors of the inserts were cleaned with wet cotton swabs. Cells were stained with a cell stain solution (0.1% crystal violet, 20% methanol) and visualized under an inverted microscope (2009 magnification; Olympus Corporation, Hachioji, Tokyo, Japan). Images were then counted using an ANALYTICAL IMAGING STATION software package (Imaging Research, Ontario, Canada).

PDAC cell proliferation assay
HPaSteC (2 9 10 4 ) were seeded in the upper well of a 24-well transwell chamber (PET membrane, 0.4-lM pore, SPL 36224, Korea) and were treated with 5 lgÁmL À1 sEVs (sEVs from HPDE, HPNE, BXPC-3, PC080, and PC084 cells) 1 day prior to co-culture. For co-culture, PDAC cells (BXPC-3, PC080, and PC084 cells; 1 9 10 4 ) were seeded in the lower well of the transwell chamber. Stellate cell medium was used as the growth medium in both compartments. After 48-h of incubation, cancer cells in the lower well were fixed in 70% (v/v) ethanol for 60 min at 4°C, cells were then stained with the cell stain solution (0.1% crystal violet, 20% methanol) and observed under the light microscope. Images were then counted using an ANA-LYTICAL IMAGING STATION software package (Imaging Research).

Soft agar assay
The soft agar colony formation assay was performed by seeding PDAC cells (5 9 10 4 ) and HPaSteC (1 9 10 5 ) in a layer of 0.4% agar/complete growth medium over the other layer of 0.8% agar/complete growth medium in the wells of a 12-well plate. Cultures were maintained in a humidified 37°C incubator. On day 5 after seeding, the colony-forming efficiency was quantified under a light microscope. Colonies whose size was quantified by the diameter of each colony were counted and analyzed if larger than 50 lm.

Animal studies
Animal experiments were approved by the Institutional Animal Care and Utilization Committee of Academia Sinica, Taipei, Taiwan (IACUC:14-05-709). Mice were maintained in a specific pathogen-free animal facility at 20 AE 2°C with a 12/12 h light/dark cycle and had free access to water and standard laboratory chow diet. All animals were monitored for abnormal tissue growth or ill effects according to the AAALAS guidelines and euthanized if the excessive deterioration of health was observed. To assess the effect of the loss of sEVs on the reprogramming of fibroblasts in the tumor, we implanted either 5 9 10 2 Rab27a/b knockdown or 1 9 10 2 WT Pan18 cells into the pancreas of 6-week-old C57BL/6JNarl (B6) mice (from Taiwan National Laboratory Animal Center). On day 30, all mice were euthanized and xenograft tumors were collected. To study the role of human PDAC cell-derived sEVs in activating fibroblasts, 6-week NOD/SCID/ IL2Rc null (NSG) male mice (Genomic Research Center, Academia Sinica, Taipei) were retro-orbitally injected with sEVs (corresponding to a total protein content of 5 lg) daily of 1 week. Next, 5 9 10 5 or 1 9 10 6 GFP-LUC-tagged PC080 cancer cells, which were resuspended in 25 lL Matrigel (Corning Inc., Corning, NY, USA), were injected orthotopically into the pancreas. After cancer cell injection, the mice were injected with sEVs (corresponding to a total protein content of 5 lg) every alternate day for 1 week. After 21 days, all mice were euthanized and xenograft tumors were collected. To study liver metastasis, intra-splenic injection of 5 9 10 4 GFP-LUC-tagged Pan18 and 1 9 10 5 mouse fibroblast cells were resuspended in 100 lL PBS and injected at the inferior pole of the spleen of 6-week-old mice using a 29G needle. After 14 days, the weight and size of the pancreatic tumor in each mouse were analyzed and liver metastases were assessed using bioluminescence IVIS imaging (PerkinElmer).

Masson trichrome staining
Collagen staining was performed using the Trichrome Stain Kit (Connective Tissue Stain; ab150686, Abcam) according to the manufacturer's protocol. To quantitatively evaluate trichrome-stained fibers in each group, representative slides per mouse were chosen, and at least five light microscopy images (409 magnification) were obtained from each slide; lastly, the percentages of stained areas were calculated using LEICA APERIO POS-ITIVE software (Aperio ImageScope v12.3).

Statistical analysis
Except for the animal studies, all experiments were repeated at least 2 or 3 times (indicated as n or N). Data were presented as the mean AE standard deviation (SD) from repeated independent experiments. Differences between various treatment groups were assessed using the Student's t test. Between-group differences were considered significant at P < 0.05. Data analyses were performed using GRAPHPAD PRISM Ver. 5.02 (San Diego, CA, USA).

Fibroblasts activated by PDAC cell-derived sEVs increase proliferation, invasion, and colony formation of PDAC
To address the potential biological effects of activated fibroblasts on pancreatic cancer cells, we analyzed PDAC cell proliferation by co-culturing human PDAC cells (BXPC-3, PC080, or PC084) and HPaSteC, which were pretreated with PDAC cell-derived sEVs or nontransformed pancreatic cell (HPDE or HPNE)-derived sEVs for 2 days. On day 2, we found that HPaSteCs treated with PDAC cell-derived sEVs instead of nontransformed pancreatic cell-derived sEVs increased PDAC cell proliferation ( Fig. 2A). In addition, HPaS-teC treated with PDAC cell-derived sEVs also significantly triggered the formation of soft agar colonies and invasion of BxPC-3, PC080, and PC084 cells (Fig. 2B,C). Similarly, primary CAFs isolated from human PDAC tissues and treated with PDAC cell-derived sEVs also promoted the formation of soft agar colonies of PC080 cells (Fig. 2D). These results demonstrated that fibroblasts activated by PDAC cellderived sEVs increased PDAC progression in in vitro models.

Knockdown of PDAC cell-derived sEV-EZR inhibits fibroblast activation
In our previous study, we found abundant expression of ezrin (EZR) in sEVs extracted from pancreatic cancer cell lines and in the plasma of PDAC [38]. EZR knockdown in sEVs derived from BXPC-3, PC080, and PC084 cells was performed using EZR shRNA and confirmed by western blotting (Fig. S2A). We found a decrease in a-SMA(+) and PDGFRB(+) fibroblasts in orthotropic tumors treated with PC080-shEZR-derived sEVs compared with those treated with PC080-shLacz-derived sEVs (Fig. 3A). Western blotting showed a decrease in a-SMA and PDGFRB expression in HPaSteCs treated with PDAC-shEZR cell-derived sEVs, compared with HPaSteCs treated with PDAC-shLacz-derived sEVs (Fig. 3B). Conversely, PDAC-EZR OE (EZR expression vector)derived sEVs increased the expression of a-SMA and PDGFRB in HPaSteCs (Fig. S2B,C). Furthermore, Transwell assays and RT-qPCR analysis showed that HPaSteC treated with PDAC-shEZR cell-derived sEVs suppressed HPaSteC migration and decreased the expression of COL1A1, COL1A2, COL3A1, and COL12A1 in fibroblasts, compared with those treated with PDAC-shLacz-derived sEVs (Fig. 3C,D). These results demonstrate that PDAC cell-derived sEV-EZR activated fibroblasts and upregulated the expression of a-SMA, PDGFRB, and the extracellular matrix in fibroblasts.

Fibroblasts activated by PDAC cell-derived sEV-EZR exacerbate PDAC metastasis
Next, we examined whether fibroblasts treated with PDAC cell-derived sEVs would affect tumor growth and metastasis of PDAC via sEV-EZR. PDAC cells co-cultured with fibroblasts treated with PDAC-shEZR cell-derived sEVs showed significantly lower colony formation than those treated with PDAC-shLacz cell-derived sEVs (Fig. 4A). In animal experiments, we injected 5 9 10 4 Pan18 GFP-luc cells and 1 9 10 5 mouse fibroblast cells into the exteriorized spleens of 6-week-old male C57BL/6J mice and monitored tumor colonization by bioluminescence imaging. Approximately 14 days postinjection, the mice developed metastatic liver tumors derived from the injected PDAC cells (Fig. 4B). The amount of liver metastasis formed by Pan18 and fibroblasts treated with Pan18-shEZR-derived sEVs was less than that formed by Pan18 and fibroblasts treated with Pan18-shLacz (Fig. 4C,D). These data suggest that fibroblasts treated with PDAC cell-derived sEV-EZR promote tumor growth and metastasis.

Discussion
Fibroblasts play a significant role in modulating tumor growth and metastasis in the tumor microenvironment. CAF infiltration is associated with not only lymphatic invasion and lymph node metastasis but also poor outcomes and shorter survival of PDAC [11,12,14,41]. The exact origin of CAFs and the mechanism by which quiescent fibroblasts develop into CAFs remain unclear. However, based on current evidence, a considerable number of CAFs originate from PSC surrounding cancer cells in PDAC [4]. PSCs are resident cells of the pancreas and quiescent in healthy individuals [42]. In chronic pancreatitis or PDAC, PSCs are activated and contribute to inflammatory and pro-fibrogenic reactions [42]. Activated PSCs express high levels of Fig. 3. PDAC cell-derived sEV-EZR activates fibroblasts. (A) Schematic illustration of animal study setup and time course. PDAC cell-derived sEVs (PC080-shLacz sEVs/shEZR-sEVs) were administered every day for 1 week before PC080 cells injection in NSG mice, followed by administration on every other day until day 21. Representative photomicrographs and the dot chart of Trichrome staining and IHC staining of a-SMA, FAP, PDGFRA, and PDGFRB positive fibroblasts in mouse pancreatic cancer tissues. Each dot represents the datum of a single mouse. Scale bar, 200 lm. 409 magnification. Data presented as mean AE SD. Level of significance was determined using the Student's t test (*P < 0.05). N = 1. (B) Results of western blotting showing expressions of EZR, a-SMA, FAP, PDGFRA, and PDGFRB in HPaSteC treated with 5 lgÁmL À1 sEVs derived from BXPC-3, PC080, or PC084 with shLacz sEVs/shEZR-sEVs for 48 h. GAPDH was used as a protein loading control. Numerical values for protein band intensities are shown below the gels. The values were quantitated by densitometry and normalized to GAPDH. N = 2. (C) Representative images and the quantified bar chart of Transwell migration assays of HPaSteC treated with 5 lgÁmL À1 sEV derived from PC080 or PC084 shLacz sEVs/shEZR-sEVs for 24 h. Scale bar, 200 lm. Values are mean AE SD (n = 3). N = 2. Level of significance was determined using the Student's t test (**P < 0.01 and ***P < 0.001). (D) qRT-PCR analysis of Col1a1, Col1a2, Col3a1, and Cal12a1 in HPaSteC treated with 5 lgÁmL À1 sEVs derived from BXPC-3, PC080 or PC084 with shLacz sEVs/shEZR-sEVs for 48 h. GAPDH was used as an internal control. Values are mean AE SD (n = 3). N = 2. Level of significance was determined using the Student's t test (***P < 0.001). shEZR, short hairpin RNA ezrin; shLacz, short hairpin RNA b-galactosidase (negative control hairpin).
a-SMA and secrete excessive amounts of ECM proteins, leading to desmoplasia in chronic pancreatitis and PDAC [42,43]. CAFs modulate cancer metastasis through the synthesis and remodeling of the extracellular matrix (ECM) and the production of growth factors, which influence angiogenesis, tumor mechanics, drug access, and therapy responses [44,45]. CAFs can activate cancer cells via multiple mechanisms in a dose-dependent manner [46]. Fibroblasts can be activated through mechanical contact with cancer cells, cancer cell-derived sEVs, and/or cytokines. For example, transforming growth factor-b (TGF-b) signaling [47], Interleukin-6 (IL-6), and PDGF are known fibroblast-activating factors [48]. sEVs released by cancer cells can transfer proteins, RNA, and miRNAs to stromal fibroblasts, thereby contributing to their recruitment and activation [49,50]. Many studies have reported that sEVs released from CAFs can be internalized by cancer cells and contribute to cancer progression and metastasis [29]. Correspondingly, sEVs released by cancer cells can induce the conversion of normal fibroblasts into CAFs. Pang et al. [51] reported that PDAC cell-derived microvesicles containing miR-155, which had been taken up by recipient fibroblasts, could convert normal fibroblasts into CAFs via the downregulation of TP53INP1. In breast cancer and melanoma, cancer cell-derived sEVs containing different miRNAs have been found to convert normal fibroblasts into CAFs, thereby contributing to tumor progression [26,30]. These results indicate that sEVs play a crucial role in the activation of fibroblasts and interactions between cancer cells and fibroblasts. a-SMA and PDGFRB are both markers of activated fibroblasts that play crucial roles in desmoplasia [52]. Previous studies have also reported that a-SMA and PDGFRB expression in fibroblasts has pro-metastatic effects mostly through matrix remodeling in invasive breast cancer [53]. In this study, we showed that sEVs released from PDAC cells upregulated the expression of a-SMA, PDGFRA, PDGFRB, and FAP and induced excessive production of collagen (COL1A1, COL1A2, COL3A1, and COL12A1) in fibroblasts ( Fig. 1). Specifically, the transfer of molecular cargo from PDAC cells to stromal cells via sEVs is a key regulator of CAF activation.
sEVs released from cancer cells are packed with unique information cargo such as lipids, proteins, and nucleic acids that can be transferred to recipient cells. Regarding the mechanism whereby PDAC cell-derived sEVs regulate fibroblast activation, we found that EZR in PDAC cell-derived sEVs plays an important role in the regulation of a-SMA and PDGFRB expression in fibroblasts, and that knockdown of EZR in PDAC cellderived sEVs attenuates the expression of a-SMA and PDGFRB in fibroblasts. Ezrin is a member of the ERM (ezrin/radixin/moesin) family, which regulates membrane remodeling [54]. EZR is involved in many cellular processes, including cell adhesion, migration, determination of cell shape, cell proliferation, and morphogenesis, in response to extracellular cues [55]. Aberrant EZR expression is associated with poor prognosis and metastasis in various cancers including PDAC [56,57]. Previous studies have shown that EZR promoted pancreatic cancer cell proliferation, migration, and invasion through upregulating the FAK/AKT signaling pathway [58]. The CAF plays a vital role in the complex process of tumor metastasis [59]. Activated fibroblasts exhibit a change in morphology and can acquire smooth musclelike properties with increased contractility, motility, proliferation, and a stellate morphology [60]. Therefore, the ezrin protein is of particular interest in fibroblasts as it plays a role in regulating cell morphology through its interaction with the membrane and actin cytoskeleton. Studies have found that co-culture of CAF with the conditioned medium of BRCA1-deficient human breast cancer cells HCC1937 can increase the expression level of ezrin mRNA in CAF by 30-fold, which helps tumor cells to metastasize and invade [61]. These results suggest that ezrin plays an essential role in the CAF to promote tumor cell metastasis. In this study, we have shown that PDAC cell-derived sEVs with ezrin activate normal fibroblasts, and reciprocally, activated fibroblasts promote tumor invasion and proliferation in in vitro transwell assays and co-culture experiments. Moreover, in our animal experiments, the amount of liver metastasis of PDAC in Pan18 cells co-injected with fibroblasts treated with shLacz-derived sEVs was larger than that in Pan18 cells co-injected with fibroblasts treated with Pan18-shEZR-derived sEVs. We demonstrated that PDAC cells secrete EZR-rich sEVs to induce a-SMA and PDGFRB expression in fibroblasts, exacerbating tumor progression and metastasis in PDAC. We observed the reversed expression of PDGFRA and PDGFRB in fibroblasts treated with sEV-EZR. This is in line with the previous studies that the expression of PDGFRB generally increases, while that of PDGFRA decreases in fibroblasts [62,63]. Previous studies have shown that the a-SMA and PDGFRB expression in fibroblast had pro-metastatic effects through matrix remodeling [53]. We found that EZR in PDAC-derived sEVs plays an important role in the regulation of a-SMA and PDGFRB expression in fibroblasts, and that knockdown of EZR in PDAC-derived sEVs attenuates the expression of a-SMA and PDGFRB in fibroblasts. This indicates that sEV-EZR can regulate CAF subtypes that express both a-SMA and PDGFRB.
The molecular mechanisms by which EZR participates in tumorigenesis are not well characterized [39], nor is the role of sEV-EZR in fibroblast activation. Previous studies have shown that the phosphorylation of STAT3 and YAP is involved in stabilizing the activated phenotype of CAFs [30,64]. STAT3 is a latent transcription factor that plays a role in regulating fibroblast function. STAT3 signaling also transmits the profibrotic effects of TGF-b on fibroblasts and that targeted inhibition of JAK2 or STAT3 ameliorates fibroblast activation and fibrosis [65]. The melanoma cell-secreted sEVs could activate CAFs through the elevation of the phosphorylation levels of JAK2 and STAT3 in CAFs [30]. Furthermore, YAP-1 overexpression augmented the proliferation, migration, expression of collagen, fibronectin, and Hippo/YAP in fibroblasts. Similarly, YAP-1 enhanced the proliferation, migration, and transition of cultured lung fibroblasts into myofibroblasts [66]. Ezrin has a direct role in the regulation of Yap in skin fibroblast proliferation. Further, ezrin depletion was shown to impair skin fibroblast proliferation through its effect on YAP nuclear translocation [40]. These results show that STAT3 and YAP are key transcriptional factors in regulating fibroblast activation and function. Our study further demonstrated that sEV-EZR regulates a-SMA and PDGFRB expression in fibroblasts via the STAT3 or YAP-1 signaling pathways. Moreover, we found that silencing of STAT3 and YAP-1 also blocks EZR-induced a-SMA and PDGFRB expression and the production of excessive amounts of collagen from fibroblasts. Lastly, we report that the YAP-1 inhibitor (CA3)/siYAP-1 could cause a greater decrease in both a-SMA and PDGFRB expression in EZRoverexpressing fibroblasts compared with the STAT3 inhibitor (Stattic)/siSTAT3. These findings suggest that YAP-1 may play a more important role in activating a-SMA and PDGFRB expression in fibroblasts. More importantly, our results confirmed that PDAC cellderived sEV-EZR modulates the STAT3 and YAP-1 signaling pathways to promote fibroblast activation, and the activated fibroblasts reciprocally exacerbate PDAC metastasis. Inhibition of EZR/STAT3 or EZR/ YAP-1 signaling pathways abrogates EZR-induced activation of fibroblasts, and targeting CAFs and desmoplasia might be a promising approach to treat  3). Level of significance was determined using the Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001). GAPDH was used as control. Numerical values for protein band intensities are shown below the gels. The values were quantitated by densitometry and normalized to GAPDH.
PDAC. Therefore, our study provides evidence that targeting STAT3 or YAP-1 or EZR inhibitors is a feasible treatment option for PDAC. Although previous studies have shown that fibroblast recruitment is essential for tumor development and that sEVs from PDAC cells can promote PSC recruitment by activating the Lin28B/let-7/HMGA2/PDGFB signaling pathway [31,67,68], further studies are needed to elucidate the mechanisms by which PDAC cell-derived sEVs recruit fibroblasts to the tumor.

Conclusion
This is the first study to show that PDAC cell-derived sEV-EZR can activate fibroblasts to express a-SMA and PDGFRB possibly via STAT3 and YAP-1 signaling pathways, which in turn exacerbate PDAC metastasis. Whether STAT3 and YAP-1 activation and sEV-EZR are linked remains to be experimentally verified. Thus, targeting sEV-EZR may contribute to improve adjuvant PDAC therapies. Fig. 6. Inhibition of EZR/STAT3 or EZR/YAP-1 signaling pathways abrogates the EZR-induced activation of Fibroblasts. (A) qRT-PCR analysis of expressions of Col1a1, Col1a2, Col3a1, and Cal12a1 in HPaSteC, whose EZR was overexpressed (transfected with EZR OE or treated with Lip-EZR) and then transfected with siRNA-STAT3 or treated with 5 lM Stattic. GAPDH was used as an internal control. Values are mean AE SD (n = 3). N = 2. Level of significance was determined using the Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001). (B) qRT-PCR analysis of the expressions of Col1a1, Col1a2, Col3a1, and Cal12a1 in HPaSteC whose EZR was overexpressed (transfected with EZR OE or treated with Lip-EZR) and then transfected with siRNA-YAP-1 or treated with 100 nM CA3. GAPDH was used as an internal control. Values are mean AE SD (n = 3). N = 2. Level of significance was determined using the Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001). (C) Representative images and the quantified bar chart of soft agar colony formation of PDAC cells co-cultured with HPaSteC treated with Lip-EZR and then transfected with siRNA-STAT3 or treated with 5 lM Stattic for 5 days. Scale bar, 100 lm. Values are mean AE SD (n = 3). N = 2. Level of significance was determined using the Student's t test (***P < 0.001). (D) Representative images and the quantified bar chart of soft agar colony formation of PDAC cells were co-cultured with HPaSteC treated with Lip-EZR and then transfected with siRNA-YAP-1 or treated with 5 lM CA3 for 5 days. Scale bar, 100 lm. Values are mean AE SD (n = 3). N = 2. Level of significance was determined using the Student's t test (***P < 0.001).

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Separation and characterization of sEVs from conditioned medium via ultracentrifugation (UC) and a sucrose density gradient (SDG). Fig. S2. sEV-EZR regulates a-SMA and PDGFRB expression in fibroblasts. Table S1. List of antibody information/resource. Table S2. List of siRNA information/resource. Table S3. List of shRNA information/resource. Table S4. List of chemical information/resource.