A total of 249 pretreatment, formalin-fixed paraffin-embedded (FFPE), EBV-positive NPC samples were collected from Linkou Chang Gung Memorial Hospital and subjected to IHC analysis. The study data were obtained through the medical records compiled by the doctors and the nursing staff and also the cancer registry, phone interviews and follow-up via correspondence with their homes or the household registration offices. A full course of therapy for NPC was defined as receiving external irradiation with a dosage of at least 68.4 Gy (Gray). The median dose of external radiotherapy was 72 Gy (68.4–82 Gy). The tumor node-metastasis (TNM) stage was defined according to the 2010 cancer staging system revised by the American Joint Committee on Cancer. All NPC tumors were histologically confirmed by pathologists. Patients were followed-up at 2- to 3-month intervals during the first 3 years after therapy and at 6-month intervals thereafter. The minimal follow-up period was 28 months. The time to local recurrence or distant metastasis was calculated using the date on which local recurrence or distant metastasis status was detected as the end point. Patients who died without occurrence of local recurrence or distant metastasis were censored in the analyses of local recurrence or distant metastasis free survival. Baseline characteristics of the YAP1 study population are summarized in Supplementary Table S1, and the characteristics of the FAPα study participants are summarized in Supplementary Table S2. Written informed consent was obtained from all participants prior to inclusion in the study.
Primary fibroblasts were isolated from nasopharyngeal biopsies and maintained in DMEM (Hyclone) supplemented with 15% fetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin (Corning). A total of 5 primary fibroblast strains (Fibro#2, Fibro#10, Fibro#17, Fibro#18, and Fibro#20) were used in this study. The nasopharyngeal cell lines, HK1 and HK1EBV, as well as rAkata cells were grown in RPMI-1640 medium (Hyclone) supplemented with 10% FBS. NPC-TW06 cells  were grown in DMEM supplemented with 10% FBS. The authenticity of HK1, HK1EBV, NPC-TW06, and rAkata cells was confirmed using short tandem repeat (STR) profile analysis (conducted by the Bioresource Collection and Research Center, Taiwan).
Non-obese diabetic (NOD)/severe-combined immunodeficient (SCID) mice were purchased from BioLASCO (Taiwan). HK1EBV cells alone (1 × 106 cells/100 μl) or together with primary fibroblasts (105 cells/100 μl), with or without pretreatment with exosomes (15 μg/ml) for 3 hours, were subcutaneously injected into the thighs of 7-week-old male NOD/SCID mice. Tumor growth was evaluated by measuring tumor size with a caliper twice a week. The tumor volume was estimated by the formula: volume (mm3) = (L × W2) / 2, where L and W are the length and width, respectively. Mice were maintained under specific pathogen-free conditions and were sacrificed when tumor volumes reached approximately 1 cm3, at which point tumor tissues were harvested for IHC analyses. All animal experiments were conducted according to accepted principles of laboratory animal care and were approved by the animal committees of National Central University, Taoyuan, Taiwan.
Total RNA from control (untreated) primary fibroblasts (Fibro#2) and fibroblasts treated with HK1EBV cell-derived exosomes (10 μg/ml) for 3 or 6 hours were extracted using an RNeasy Mini kit (Qiagen). cDNA libraries were constructed and sequenced on the Illumina NovaSeq platform. The cutoff value for differentially expressed genes (DEG) was defined as absolute log2 fold change > 0.5 (equal to a 1.42-fold change in the original scaling) and p < 0.05. Functional analyses, including Gene Set Variation Analysis (GSVA)  and KEGG pathway enrichment analysis, were performed using R package clusterProfiler , GSEA  and Ingenuity Pathway Analysis software (QIAGEN) based on statistical significance and DEG values.
Collagen gel contraction assay
Fibroblast cell suspension was mixed with pH-adjusted collagen solution (Advanced Biomatrix) at a final collagen concentration of 1.98 mg/ml. A 500-μl sample of this cell-collagen mixture containing 3 × 104 cells was seeded into wells pre-coated with bovine serum albumin (BSA) and incubated at 37 °C for 90 minutes until the gel solidified. Polymerized cell-collagen mixtures were then overlaid with 500 μl of DMEM supplemented with 5% FBS plus exosomes or inhibitors. After a 1-hour incubation, the edge of the gel was carefully dislodged with a 20-μl tip, and the area of gel contraction was measured using ImageJ software. All contractility experiments were conducted in triplicate, and at least three independent experiments were conducted.
IHC analysis of FFPE tumor tissue sections
Pretreatment NPC FFPE biopsy sections were collected for examination of the expression of relevant proteins. Primary antibodies against the following proteins were used for IHC analyses: human FAPα (Abcam, ab207178), YAP1 (Cell Signaling Technology, #14074), active YAP1 (Abcam, ab205270), PDGFRα/β (Abcam, ab32570), CD8a (Cell Signaling Technology, #85336), and granzyme B (Cell Signaling Technology, #46890). The fibrotic response of the tumor was evaluated using trichrome staining (ScyTek). The IHC staining score and degree of fibrosis were defined according to staining intensity (“−”, “+”, “++” or “+++”) and extent (positive percentage relative to total area investigated). All biopsies were histologically confirmed by pathologists.
Cells and exosomes were lysed in NP40 lysis buffer containing a protease inhibitor cocktail (Roche). Equal amounts of protein were loaded and separated by sodium dodecyl sulfate-polyacrylamide electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride membranes (GE Healthcare Life Science). The membranes were washed briefly with 0.1% Tween-20 in TBS (TBS-T), blocked with 5% BSA in TBS-T, and probed with specific primary antibodies at 4 °C overnight. After serial washes with TBS-T, the blots were incubated with horseradish peroxidase (HRP)-conjugated secondary antibody (Millipore), followed by development with chemiluminescence reagents.
Isolation of exosomes
EBV-positive HK1EBV, rAkata, or HK1 cells were cultured overnight in complete RPMI-1640 at a density of 2 × 106 cells/ml after which the medium was replaced with 10 ml of RPMI-1640 medium containing 1% EV-depleted FBS and cells were cultured for an additional 48 hours. The cell-conditioned supernatant was collected for total exosome isolation using ExoQuick-TC according the manufacturer’s instructions (System Biosciences). Isolated exosome pellets were lysed in 1% NP-40 lysis buffer for Western blot analysis or resuspended in an appropriate amount of PBS and stored at − 80 °C until use.
TEM analysis of exosomes
A mixture of freshly prepared exosomes and magnetic beads (Invitrogen Dynabeads M− 270 Epoxy) conjugated to anti-CD9 antibody (BD Pharmingen, 555,370) was incubated for 30 minutes and then centrifuged for 1 minute, yielding exosome-captured magnetic beads. The precipitated beads were blocked with 1% BSA/PBS, washed with PBS, and resuspended in PBS. Five microliters of this captured exosome solution was loaded onto a 200-mesh copper grid (TED PELLA INC), incubated for 8 minutes at room temperature, and washed with distilled water to remove excess sample. For negative staining, 1.5% (w/v) phosphotungstic acid hydrate solution (Merck, Darmstadt, Germany) was applied to the copper grid for 3 seconds followed by washing with distilled water. Exosomes were examined under a transmission electron microscope operating at 200 kV (JEM2000FX, JEOL).
Exosomal stimulation of primary fibroblasts
Human fibroblasts were cultured in fresh DMEM supplemented with 3% FBS for 16 hours, and then treated with 10 μg/ml exosomes at 37 °C or varying durations, as indicated. Uptake of exosomes was evaluated by imaging analysis using ExoGlow (green)-labeled exosomes (System Biosciences) or detection of exosome-specific markers.
Preparation of conditioned medium
Primary fibroblasts were plated at a density of 1 × 105 cells/well in 6-well plates and grown for 1 day. The culture medium was replaced with DMEM supplemented with 3% FBS and grown at 37 °C for an additional 16 hours. Cells were then treated with exosomes or as described in the text. Cell-free culture supernatants were collected and centrifuged at 2000 rpm, 4 °C for 10 minutes. Conditioned medium was stored at − 80 °C until use. Verteporfin (1 μM) or saracatinib (20 μM) was added to the growth medium of fibroblasts 2 hours prior to exosome (10 μg/ml) treatment.
Real-time monitoring of cellular responses using the xCelligence system
The xCelligence real-time analyzer (ACEA Biosciences) was utilized to monitor the growth and survival of cells based on measurements of cell impedance. Prior to cell seeding, the background impedance of each E-plate was determined by loading 50 μl/well of culture medium. Cells (3 × 103 cells/well in 100 μl) were then plated and allowed to grow at 37 °C. Impedance was recorded every 10 minutes for 3 to 5 days. Cytokines [IL6 (20 ng/ml; R&D Systems), IL8 (50 ng/ml; R&D Systems), CCL2 (100 ng/ml; R&D Systems)], cytokine-specific neutralizing antibodies [IL6 (0.5 μg/ml; Abcam, Ab9324), IL8 (0.5 μg/ml; Cell Signaling, #94407), and CCL2 (0.5 μg/ml; Invitrogen, MA5–17040)], the corresponding IgG isotype control antibodies (0.5 μg/ml) (Croyez Bioscience), inhibitors (verteporfin and saracatinib), or conditioned medium was added at the indicated time points. All experiments were performed in quadruplicate.
5-Ethynyl-2-deoxyuridine (EdU) incorporation assay
DNA synthesis in cells was evaluated using a Click-iT EdU Alexa Fluor 488 EdU Incorporation kit (Invitrogen) according to the manufacturer’s protocol. Briefly, NPC cells were treated with conditioned medium or cytokines for 10 hours, followed by EdU (10 μM) labeling for 4 hours. Cells were then fixed and counterstained with Hoechst 33342 DNA dye. The number of EdU-positive cells was determined using cellSens Software (Olympus).
For live-cell imaging of the actin cytoskeleton, human primary fibroblasts were infected with adenovirus containing LifeAct-RFP (Ibidi) at a pre-optimized multiplicity of infection (MOI) according to the manufacturer’s protocol. For selection of stable clones, puromycin (2 μg/ml) (Invitrogen) was added to culture medium 48 hours after infection. Time-lapse experiments were performed in a microscope chamber maintained at 37°Cand 5% CO2 (Olympus IX830). Phase-contrast and fluorescent images were simultaneously acquired every 2 minutes, and movies were obtained using CellSens imaging software (Olympus).
Quantitative real-time reverse transcription-polymerase chain reaction (qRT-PCR) analysis
An optimal amount of total RNA was reverse transcribed into cDNA using a first strand cDNA synthesis kit (Roche) according to the manufacturer’s instructions. Diluted cDNAs were mixed with Master mix reagents (Bio-Rad), and PCR was performed on a LightCycler 96 Instrument (Roche) following the manufacturer’s protocol. Primer information is summarized in Supplementary Table S3. All assays were repeated in duplicate.
T cell cytotoxicity assay of NPC spheroids
Human peripheral blood-derived T cells were expanded and activated by using recombinant IL-2 (30 U/ml) (R&D Systems) and Dynabeads magnetic beads coated with anti-CD3 and anti-CD28 antibodies (Gibco). A 3-dimensional spheroid coculture model was established for tumor killing assays by activated T cells. Cancer spheroids were generated according to the manufacturer’s instructions (STEMCELL Technologies). Briefly, NPC-TW06-GFP cells (9 × 105 cells/well) alone or mixed with human fibroblasts (2 × 105 cells/well) were added to the wells containing microwells of a 24-well AggreWell plate (STEMCELL Technologies) and centrifuged to evenly distribute cells in the microwells. Spheroids were formed after 24 hours of incubation. Four-day activated T cells were then added to spheroid cultures with an effector/target ratio of 3:1 and cultured for another 24 hours. T cell-mediated tumor killing was evaluated via microscopic examination (Olympus IX 83), and the integrated GFP intensity of spheroids was calculated using ImageJ software.
Reagents and antibodies
Stock solutions of saracatinib (LC Laboratories) and verteporfin (AdooQ Bioscience) were prepared by dissolving in DMSO at final concentrations of 369 and 69 mM, respectively. Unless otherwise indicated, saracatinib and verteporfin were used at final concentrations of 20 μM and 1 μM, respectively, in all drug treatment experiments. The human recombinant cytokines, IL6 (20 ng/ml; R&D Systems), IL8 (50 ng/ml; R&D Systems) and CCL2 (100 ng/ml; R&D Systems), were used to investigate effects of cytokines on cell proliferation and activation of signaling pathways in NPC cells. Human ELISA kits (R&D Systems) were used to detect levels of IL6, IL8, and CCL2 secreted into the culture medium of fibroblasts. Treatment of Brefeldin A (BFA; BD Biosciences, 555,029) was used to increase the detectability of intracellular cytokines by Western blotting. The indicated antibodies against the following proteins were used for Western blotting: calnexin (Cell Signaling, #2679), phospho-YAP (Cell Signaling, #13008), YAP (Cell Signaling, #14074), CYR61 (Cell Signaling, #14479), CTGF (Cell Signaling, #86641), IGFBP3 (Cell Signaling, #25864), PDGFRα/β (Abcam, ab32570), α-SMA (Abcam, ab124964), phospho-mTOR (Cell Signaling, #2971), phospho-p70S6K1 (Cell Signaling, #9234.), phospho-MEK1/2 (Cell Signaling, #9121), MEK1/2 (Cell Signaling, #4694), phospho-p38 MAPK (Cell Signaling, #4511), p38 MAPK (Cell Signaling, #9212), phospho-ERK1/2 (Cell Signaling, #4370), and ERK1/2 (Cell Signaling, #4695), CD9 (Abcam, ab92726), GAPDH (Abcam, ab9484), active YAP1 (Abcam, ab205270), mTOR (Abcam, ab2732), p70S6K1 (Abcam, ab32529), HSP70 (System Biosciences, EXOAB-HSP70A-1), β-actin (Sigma, A5060), α-tubulin (Santa Cruz Biotechnology, sc32293), and EBV-encoded LMP1 monoclonal antibody (S12), purified from hybridoma culture supernatants. Uncut western blots for all figures were summarized in Supplementary Fig. S1. Primary antibodies against the following proteins were used for immunocytochemistry: FAPα (Santa Cruz Biotechnology, sc65398,), active YAP1 (Abcam, ab205270), CYR61 (Cell Signaling, #14479), IGFBP3 (Cell Signaling, #64143), and CTGF (Cell Signaling, #86641).
Statistical analyses were performed using SPSS 16.0 (SPSS) or GraphPad Prism 9 (GraphPad Software). The statistical significance of immunoreactivities in human biopsies was assessed using the chi-square test. Spearman’s rank correlation coefficient was used to evaluate correlations between IHC results (staining intensity × percentage). Kaplan-Meier survival and log-rank tests were used to compare survival times between groups based on IHC scores. Univariate and multivariate Cox regression models were used to identify independent prognostic factors for metastasis-free, recurrence-free, and overall survival. Variables tested in univariate and multivariate analyses included YAP1 or FAPα score, treatment days, biologically effective dose, age, gender, T stage, N stage, chemotherapy, comorbidity, smoking status, betel quid, and alcohol consumption. All statistical tests were two-sided, and p-values < 0.05 were considered statistically significant.