Exosomes from HNSCC Promote Angiogenesis through Reprogramming of Endothelial Cells

This article is featured in Highlights of This Issue, p. 1615

Head and neck squamous cell carcinomas (HNSCC) are a heterogenous group of carcinomas that affect the oral cavity, the pharynx, and the larynx. They represent the largest proportion of all malignant tumors in the head and neck region and are often associated with chronic tobacco consumption and alcohol abuse (1). The infection with human papillomavirus (HPV) also appears to be an etiologic factor for HNSCC, and HPV-positive (HPV⁺) and HPV-negative (HPV⁻) tumors are considered to represent two different clinicopathologic and molecular entities (2, 3). For the growth of solid tumors such as HNSCC, an adequate autonomous blood supply is essential to supply the neoplastic tissue with nutrients and oxygen (4). Tumor angiogenesis is regulated by a complex network of morphogenetic and molecular pathways as well as soluble factors (5). Recently, tumor-derived exosomes (TEX), virus-size vesicles, which circulate freely throughout body fluids and accumulate in the tumor microenvironment (TME), have been recognized as a new contributor to angiogenesis.

Exosomes are produced by all cells and serve as a universal communication system (6). Tumor cells are active exosome producers, and the analysis of plasma of cancer patients shows greatly increased exosome numbers compared with plasma of healthy blood donors, especially for patients with HNSCC with advanced or metastatic disease (7). In the TME, exosomes mediate autocrine, juxtacrine, and paracrine interactions (8). Their effects on the blood vessel establishment and development are being intensively investigated (9). The cell line- or plasma-derived exosomes from a variety of human tumors, including, for example, glioblastoma or pancreatic and nasopharyngeal carcinomas, were found to be potent inducers of angiogenesis in vitro and in vivo (10–12). The TEX cargo contains a variety of angiogenesis-related proteins, including VEGF, TGFβ, bFGF, MMP2, and MMP9 (10, 13). TEX have been reported to induce phenotypic and functional changes in endothelial cells (EC). However, little is known about the mechanisms underlying the induction of tumor angiogenesis by TEX. Potentially, the transfer of RNA, miRNA, and proteins to ECs contributes to the initiation and progression of tumor angiogenesis (14). Those effects can be enhanced by environmental factors such as hypoxia, which can lead to substantial alterations in the molecular/genetic content of TEX, which appear to be especially effective in reprogramming EC characteristics and in promoting angiogenesis (15).

To date, few studies have examined TEX in HNSCC. However, it is known that cultured HNSCC cells produce TEX that can be isolated by a mini-SEC method in a biologically active form (16). Exosomes have also been detected in the blood of patients with HNSCC and shown to be enriched in numerous immunosuppressive proteins (7). To what extent TEX produced by HNSCC cells contribute to the induction of angiogenesis has not yet been reported. Here, we evaluate the role TEX produced by HNSCC cells play in blood vessel formation in vitro using as targets human umbilical vein endothelial cells (HUVEC) and in vivo using an immunocompetent orthotopic mouse model of oral cancer.

The HPV⁻ cell line PCI-13 was established and maintained in our laboratory (17) and the HPV⁺ cell lines UMSCC47 and UMSCC90 were established by Dr. Thomas Carey (University of Michigan, Ann Arbor, MI). UMSCC47, UMSCC90, and the murine head and neck carcinoma cell line SCCVII were obtained from Robert L. Ferris (UPMC Hillman Cancer Center, Pittsburgh, PA). All cell lines were authenticated prior to their use. Cells were grown in DMEM (Lonza Inc.) supplemented with 1% (v/v) penicillin/streptomycin and 10% (v/v) exosome-depleted and heat-inactivated FBS (Gibco, Thermo Fisher Scientific) at 37°C and in the atmosphere of 5% CO₂ in air. For certain experiments, cells were exposed to hypoxic conditions and were cultured in an Heracell 150i CO₂ incubator (Thermo Fisher Scientific) in a humidified atmosphere of 1% O₂ and 5% CO₂ at 37°C. For exosome isolation, 4 × 10⁶ PCI-13 cells, 2.5 × 10⁶ UMSCC47 cells, 5 × 10⁶ UMSCC90 cells, or 4 × 10⁶ SCCVII cells were seeded with 25 mL media in 150 cm² cell culture flasks. Supernatants were collected after 72 hours. HUVECs were obtained from Cell Systems and were cultured in medium 200 with 10% (v/v) Low Serum Growth Supplement (Gibco).

TEX were isolated by using the mini-SEC method as described previously (16). Briefly, cell culture supernatants were centrifuged at room temperature for 10 minutes at 2,000 × g and then at 4°C for 30 minutes at 10,000 × g followed by ultrafiltration using a 0.22-μm filter (EMD Millipore). Aliquots (50 mL) of supernatants were concentrated to 1 mL using Vivacell 100 concentrators (100,000 MWCO, Sartorius) at 2,000 × g, and 1 mL of the concentrate was placed on a mini-SEC column. TEX were eluted 1-mL fractions using PBS. TEX in fraction #4 were harvested as described previously (16). Protein concentrations were measured using a BCA protein assay (Pierce Biotechnology). To concentrate isolated TEX, 0.5 mL 100K Amicon Ultra centrifugal filters (EMD Millipore) were used for centrifugation at 2,000 × g.

Peripheral venous blood specimens were collected from 10 patients with HNSCC with active disease (AD) seen at the UPMC Otolaryngology Clinic in 2016 or 2017. Blood specimen from 3 normal donors (ND) served as control. Informed consent from all individuals was obtained and the study was approved by the institutional review board of the University of Pittsburgh (UPCI 09-069/IRB991206). The blood samples were delivered to the laboratory and were centrifuged at 1,000 × g for 10 minutes to separate the plasma from blood components. Plasma specimens were stored in 1-mL aliquots at −80°C and were thawed immediately prior exosome isolation. Thawed plasma samples were differentially centrifuged at 2,000 × g for 10 minutes at room temperature and at 10,000 × g for 30 minutes at 4°C. After ultrafiltration using a 0.22-μm filter (EMD Millipore), the exosome isolation from 1 mL of plasma was performed by the mini-SEC method as described above.

Transmission electron microscopy (TEM) was performed as described previously (7) at the Center for Biologic Imaging, the University of Pittsburgh. Freshly isolated TEX were placed on a copper grid coated with 0.125% Formvar in chloroform and stained with 1% (v/v) uranyl acetate in ddH₂O. A JEM 1011 microscope was used for exosome visualization.

Size distribution and concentrations of the particles in isolated exosome fractions were analyzed using tunable-resistive pulse sensing (TRPS) by qNano (Izon) as described previously (16). All samples were measured under following conditions: NP#49607, stretch 45.64 mm, voltage 0.7 V, and two pressure steps 3 to 7 mbar. The particle calibration (Part#: CRC100b, mean diameter: 114 nm, dilution: 1:1,000) was measured directly after the experimental sample under identical conditions. Data recording and analysis were performed using the Izon software (version 3.2).

The presence of TSG101, an endocytic protein, in TEX was confirmed by Western blot analyses as described previously (7, 16, 18). Each lane was loaded with 10 μg of exosome proteins, and polyvinylidene difluoride membranes were incubated overnight at 4°C with a TSG101 antibody (1:1,000, ab30871, Abcam).

The relative levels of human angiogenesis–related proteins in TEX or their parent cell lines were measured using a Human Angiogenesis Array Kit (R&D Systems Inc.). Aliquots of TEX (200 μg protein) isolated from supernatants of PCI-13 or UMSCC47 cell lines or of the same cells lysed with Laemmli sample buffer (Bio-Rad Laboratories) were added to the array, and the results were analyzed with the ImageJ software (http://rsbweb.nih.gov/ij/).

RNA was isolated from 2 × 10⁵ HUVECs cultivated with 20 μg of PCI-13- or UMSCC47-derived exosomes in wells of 24-well plates using a miRCURY RNA Isolation Kit (Qiagen). qPCR was performed using QuantiNova SYBR Green RT-PCR Kit (Qiagen) in Fast 96-Well Reaction Plates (Applied Biosystems) using a StepOnePlus (Applied Biosystems). The following PCR primers were used: β-actin primers; forward 5′-GTGGGGCGCCCCAGGCACCA-3′, reverse 5′-CTCCTTAATGTCACGCACGATTTC-3′. VEGF primers; forward 5′-GAGAATTCGGCCTCCGAAACCATGAACTTTCTGCT-3′, reverse 5′-GAGCATGCCCTCCTGCCCGGCTCACCGC-3′. IGFBP3 primers; and forward 5′-TGGGGATATAAACAGCCCAGC-3′, reverse 5′-GGTCACCCCAGTCACTCCT-3′. Relative mRNA levels were calculated by the ΔΔC_t method using β-actin for normalization.

Isolated TEX were labeled with DiD membrane-labeling solution (Invitrogen) using manufacturer's instructions. Briefly, 500 ng/mL TEX solution was incubated with 10 μL of 5 μmol/L DiD solution for 30 minutes at 37°C to stain the exosomal membrane. Excessive dye was removed by washing 20× with PBS using 100K Amicon Ultra centrifugal filters. HUVECs were seeded on collagen type 1–coated coverslips at the density of 5,000 cells/mm². After overnight serum starvation, 10 μg/mL DiD-labeled TEX were added at the designated time points. To wash-off TEX bound to the HUVEC surface, cells were treated with stripping buffer (14.6 g NaCl, 2.5 mL acetic acid, 500 mL distilled water) for 2 minutes, followed by extensive washing. Cells were fixed with 3.33% paraformaldehyde and labeled with Alexafluor488-Phallodin (5:200) and Hoechst 33342 (1:1,000). Imaging was performed using Carl Zeiss LSM 880 confocal microscope (Carl Zeiss Microscopy) and the images were processed using ZEN Blue software (Carl Zeiss Microscopy). All experiments were performed in triplicates, and for each individual experiment, five images were taken at random locations. Imaging, processing, and analysis were done under constant settings across all the time points and relative fluorescence intensities (RFI) were calculated using ImageJ software after the background correction. To assess mechanisms responsible for internalization, cells were pretreated with different inhibitors prior to the addition of TEX. Potential cytotoxicity of inhibitors was assessed using the Calcein AM Cell Viability Assay (R&D Systems Inc.). The highest concentration of the inhibitor that was nontoxic to the cells was used in the subsequent assays. To assess the effect of inhibitors on TEX uptake, HUVECs were preincubated with pharmacologic/chemical inhibitors before exosome addition. For genistein treatment, cells were pretreated with 100 μmol/L genistein at 37°C for 30 minutes and genistein was also present throughout the experiment as described previously (19). Cells were pretreated with 50 μmol/L nystatin, 5 mmol/L MβCD, or 50 μmol/L heparin for 60 minutes at 37°C and were thoroughly washed prior to the addition of DiD-labeled TEX. The DMSO concentration was always kept below a final concentration of 0.1% (v/v). After 4 hours of incubation, cells were fixed, stained, and imaged as described above.

HUVECs (2 × 10⁴) were resuspended in serum-free media and placed on top of 70 μL growth factor–reduced Matrigel (Corning Inc.) in wells of 48-well plates. Cells were treated with 10, 20, or 50 μg of TEX per well. Following incubation at 37°C for 6 hours, tubules were imaged in 5 random regions of interest, using phase contrast microscopy at 10× magnification (Axiovert 25 CFL, Carl Zeiss Microscopy). Tubule length and numbers of branch points were analyzed with the Angiogenesis Analyzer developed for the ImageJ software (20).

HUVECs (5 × 10³) were seeded into wells of 96-well plates with 5 or 10 μg of TEX per well, and incubated for 24 or 48 hours. The MTS cell proliferation assay was performed according to the manufacturer's instructions (Abcam). Absorbance was measured at 490 nm.

Subconfluent cell monolayers in wells of 48-well plates were treated with 10 μg/well of TEX for 24 hours. After incubation, the confluent monolayer was wounded mechanically by using pipet tips and the wound was imaged using an Axiovert 25 CFL inverted microscope at 5× magnification. Imaging was repeated 6 and 12 hours after the initial scratch. Wound closure was analyzed using ImageJ and results are expressed as percent of recovery.

HUVECs (5 × 10⁴) were starved in serum-free media overnight and were added to the top compartment of 24-well transwell plates with 8-μm pore diameter (Corning Inc.). Cells migrated toward serum-free media, 10, 20, or 50 μg of PCI-13 or UMSCC47 exosomes or 10% FBS, which were added to the bottom compartment. After 6 hours of incubation at 37°C, nonmigrating cells in the top chamber were removed with cotton swabs. Migrating cells on the bottom surface of the membrane were fixed in methanol and stained with 0.2% crystal violet (Sigma-Aldrich). The number of migrated cells was counted in a light microscope in six randomly selected regions of interest at 20× magnification using an Olympus BX51 microscope (Olympus America).

Tumor cell to EC adhesion was quantified by using the CytoSelect Tumor-Endothelium Adhesion Assay (Cell Biolabs, Inc.) according to the manufacturer's instructions. Briefly, a monolayer of HUVECs was treated with TEX (10 μg) or TNFα for 12 hours. Fluorescently labeled PCI-13 or UMSCC47 cells (1 × 10⁶ cells/mL) were allowed to attach to the HUVEC monolayer for 1 hour. The monolayer was washed gently, and adherent cells were visualized at 10× magnification using an Axiovert 25 CFL inverted fluorescence microscope. Adherent cells were quantified in five random regions of interest per well using ImageJ.

Thirty-two female C57BL/6 mice aged 6 to 8 weeks were purchased from Jackson Laboratories. Protocols for animal experiments were approved by the Institutional Animal Care and Use Committee under the reference number 16088780. To induce the development of oral carcinomas, mice had 4-nitroquinoline-1-oxide (4-NQO; Sigma-Aldrich) administered in drinking water at 100 μg/mL for a 16-week period, followed by the provision of normal drinking water. During this period, tumor incidence was 100%. Mice were randomly divided into four experimental groups: 90 μg of exosomes derived from UMSCC90, PCI-13, or SCCVII cell lines were injected intravenously via the tail vein. The control group received sterile PBS (Lonza Inc.) injections. Mice were monitored 2 to 3 time/week starting from the initiation of 4-NQO treatment for weight loss or signs of reduced/altered behavior. The mice were euthanized at the point of 20% weight loss or after 27 weeks from the initial administration of 4-NQO. For the tissue histology, oropharyngeal tumors with a volume of approximately 3 mm³ were dissected, placed in 2% paraformaldehyde for 1 hour, and subsequently in 30% sucrose (Sigma-Aldrich) for 24 hours. Samples were embedded in OCT compound (Thermo Fisher Scientific) and stored at −80°C for subsequent sectioning.

Cryostat sections (6 μm) were cut and stained with hematoxylin and eosin (H&E). Immunofluorescence staining was performed by incubating sections with a rat anti-mouse CD31 Ab (1:100, clone MEC13.3, BD Biosciences) and a rabbit anti-mouse α-SMA Ab (1:100, ab5694, Abcam) overnight at 4°C. After washing, tissue sections were incubated with donkey anti-rat Alexa Fluor 488 or goat anti-rabbit Alexa Fluor 488 Ab (1:500, Invitrogen) for 1 hour at room temperature. Negative controls were stained in parallel with the secondary antibodies alone. Sections were counterstained with Hoechst nuclear stain, mounted and imaged using an Olympus BX51 microscope. Histology was evaluated blinded by quantifying green fluorescence in three random regions of interest inside the tumor tissue with ImageJ. Data are expressed as the percentage of the area that was positively stained from the region of interest (% of ROI).

All data were analyzed using GraphPad Prism (v7.0). Results were expressed as means ± SEM. Differences between groups were assessed by Student t test or by one-way ANOVA with post hoc analysis when appropriate. Differences were considered significant at P < 0.05.

The exosome #4 fractions isolated from supernatants of UMSCC47 and PCI-13 cell lines by mini-SEC contained 27.2 ± 4.9 μg proteins per 10⁵ PCI-13 cells and 43.9 ± 6.1 μg proteins per 10⁵ UMSCC47 cells in fractions #4, respectively. Culturing tumor cells under hypoxic conditions significantly increased the yield of total exosome proteins for both cell lines (P < 0.05; Fig. 1). The mean exosome sizes were 116 nm for PCI-13 and 122 nm for UMSCC47. They had a cup-shaped morphology (Supplementary Fig. S1B), and carried TSG101 (Supplementary Fig. S1C). TEX isolated using the mini-SEC method showed the typical size and vesicular morphology, and were of endocytic origin.

To investigate the presence of proangiogenic proteins in tumor cells and exosomes produced by these cells, angiogenesis arrays were used. Tumor cells as well as TEX had a broad spectrum of angiogenesis-related proteins (Fig. 2). Specifically, PCI-13–derived exosomes carried proangiogenic proteins such as uPA, coagulation factor III, and MMP-9, but also antiangiogenic proteins such as thrombospondin-1 (Fig. 2D). UMSCC47 cells and exosomes carried an even wider spectrum of angiogenesis-related proteins, such as coagulation factor III, IGFBP-3, uroplasminogen activator (uPA), thrombospondin-1, and endostatin (Fig. 2E). Coincubation for 24 hours of 20 μg PCI-13–derived exosomes with HUVECs caused an increase in VEGF mRNA levels and IGFBP-3 mRNA expression levels in the recipient cells (P < 0.05). No significant changes were seen after coincubation of HUVECs with UMSCC47-derived exosomes (Fig. 2A). The mRNA expression levels of ANGPT1, FGF2, and PDGFA were not influenced by TEX (data not shown). PCI-13 and UMSCC47 cells and deriving exosomes express a variety of angiogenic markers. Treatment of HUVECs by PCI-13–derived exosomes leads to upregulation of mRNA levels for angiogenesis-related genes.

To investigate interactions between TEX and HUVECs, TEX were stained with the DiD dye and their uptake into HUVECs was evaluated by confocal microscopy (Fig. 3A). Confirmation of exosome internalization is provided in Supplementary Movie S1. TEX uptake by HUVECs started after 15 minutes of coincubation and increased constantly over time (P < 0.05; Fig. 3B). Next, a spectrum of pharmacologic inhibitors was used to block specific internalization pathways. Cell viability was not influenced by genistein, nystatin, and heparin in the tested concentrations, but MβCD concentration had to be reduced to 5 mmol/L to retain HUVEC functionality (Supplementary Fig. S2). DMSO (0.1%) had no effects on exosome uptake. The exosome internalization was not significantly inhibited by MβCD, and inhibition by genistein and nystatin was from 30% to 50% (P < 0.01). Heparin blocked exosome internalization most effectively (70%–80%; P < 0.01; Fig. 3B and D). No significant differences were observed by comparing internalization properties of PCI-13 and UMSCC47-derived exosomes. HUVECs internalize TEX produced by head and neck cancer cells within 4 hours and internalization can be inhibited by blockage of specific uptake mechanisms.

TEX derived from PCI-13 and UMSCC47 cell lines modulated HUVECs phenotypically and led to an increased formation of vascular structures after plating on Matrigel (Fig. 4A). The tubule length was significantly increased as a response to exosome treatment in a concentration-dependent manner (P < 0.001; Fig. 4A). The number of branch points was significantly increased after treatment with 10 μg of PCI-13–derived exosomes or 20 μg of UMSCC47-derived exosomes compared with the nontreatment controls (P < 0.01). In transwell studies, HUVEC migration toward TEX was significantly inhibited within 6 hours by increasing the level of total exosomal protein (Fig. 4B). The analysis of wound healing after 12 hours of coincubation with TEX showed no significant differences after 6 hours, but was increased after 12 hours of incubation with TEX (P < 0.05; Fig. 4C). Starting the TEX treatment simultaneously to the wounding of the culture led to no significant differences after 12 hours (data not shown). In addition, proliferation of HUVECs in response to TEX significantly increased as a function of concentration and time. Proliferation was increased for TEX derived from normoxic and hypoxic cultures and for 24 or 48 hours of incubation (Fig. 4D). Low concentrations of TEX derived from normoxic cultures showed a higher ability to stimulate HUVEC proliferation. No differences between hypoxia and normoxia-derived TEX were observed at higher TEX concentrations. In functional studies, exosomes derived from HNSCC cell lines stimulated tube formation, proliferation, and migration of HUVECs and, thus induced functional reprogramming of ECs in vitro.

To evaluate whether TEX were able to stimulate tumor cell adherence to ECs, HUVEC monolayers were pretreated with PCI-13- or UMSCC47-derived exosomes for 6 hours and then PCI-13 and UMSCC47 cells were added to measure their adherence to HUVECs. The adherence of tumor cells from both cell lines was increased after the cell monolayers were treated with their own exosomes (P < 0.0001). TEX isolated from hypoxic cultures showed a lower ability to stimulate tumor cell adherence compared with TEX from normoxic cultures. UMSCC47-derived exosomes also increased PCI-13 cell adherence, albeit to a lesser extent than autologous TEX (P < 0.05). Hypoxia-derived PCI-13 exosomes led to a highly increase of UMSCC47 cell adherence (P < 0.0001). Relative to UMSCC47 adherence to HUVECs, PCI-13 cells were significantly more adherent to exosome-treated HUVECs (Fig. 5). The stimulating effects of TEX on tumor cell adherence to HUVEC monolayers were partially blocked (P < 0.01) by using the uptake inhibitor, heparin, whereas heparinization by itself had no effect on tumor cell adherence. Therefore, TEX internalization, rather than receptor–ligand interactions on the surface of the recipient cells, is necessary for reprogramming of the HUVEC monolayers. HNSCC-derived exosomes reprogram HUVEC monolayers promoting adherence of cancer cells to these monolayers.

To further investigate the promoting role of TEX in tumor angiogenesis, we isolated exosomes from the plasma of 10 patients with active disease prior to any therapy (AD) and compared them with exosomes from plasma of ND. Supplementary Table S1 provides clinicopathologic characteristics of the patients enrolled in the study. We observed higher numbers of exosomal protein in plasma from patients with UICC stages I/II (P = 0.044) and UICC stages III/IV (P = 0.0082) compared with ND plasma (Fig. 6B). Plasma-derived exosomes were rapidly internalized by HUVECs (30 minutes–4 hours). No differences in uptake were observed between AD or ND exosomes, or plasma- and cell line–derived exosomes (Fig. 6A). The proliferation of HUVECs was increased after treatment with ND or AD plasma–derived exosomes for 48 hours. Exosomes from plasma of AD patients with the UICC stages II–IV increased the HUVEC proliferation compared with exosomes from UICC stage I patients (P = 0.0044) or NDs (P = 0.0024; Fig. 6C). These data indicate that patients with HNSCC with active disease produce elevated levels of exosomes that interact with HUVECs and stimulate their proliferation.

To evaluate whether TEX derived from HNSCC cell lines stimulate angiogenesis in vivo, we used the 4-NQO orthotopic head and neck cancer mouse model and injected TEX intravenously. All mice showed premalignant lesions after 4-NQO treatment, which turned to malignant lesions in the period of 16 weeks. Oral tumors were characterized histopathologically and stained for vascular structures. H&E staining revealed similar staining patterns for all groups, showing large, partially pleomorphic, and atypical cells. Tumors disrupted the basal layer of the tongue epithelium and infiltrated the neighboring tissue. 4-NQO treatment led to the tumorigenesis of poorly differentiated squamous cell carcinomas, showing a meager keratinization (Fig. 7A). The detection of CD31, an endothelium marker, revealed an increased vascularization in the border zones of the tumors in all groups. Exosome-treated groups showed an increased amount of vascular structures within the tumor tissue. We observed increased CD31-positive signals inside the tumors of all exosome-treated groups with PCI-13 showing the highest vascularization (P < 0.05; Fig. 7B). We discovered analogous results by detecting pericytes using an α-SMA antibody. TEX treatment resulted in increased α-SMA signals compared with the untreated group, which were significantly increased after the injection of PCI-13 and SCCVII-derived exosomes (P < 0.05; Fig. 7C). In the untreated group, pericytes were mainly present on blood vessels in the border zone between tumor tissue and normal tongue tissue. The TEX treatment resulted in an increased coverage with pericytes of blood vessels within the tumors (Fig. 7A). Overall, TEX promote angiogenesis measured by detection of ECs and pericytes in vivo.

For solid tumors, an adequate blood supply is of critical importance for the development, growth, and metastasis (21). Our data indicate that TEX may be one of the underlying mechanisms responsible for the acceleration of tumor angiogenesis in HNSCC. TEX derived from HPV⁺ and HPV⁻ HNSCC cell lines accelerated in vitro and in vivo angiogenesis. A closer examination of TEX produced by PCI-13 (HPV⁻) and UMSCC47 (HPV⁺) cells revealed that they carry a variety of angiogenesis-related proteins, similar to exosomes from glioblastoma (10, 13). The protein content of TEX likely drives the progression of tumor angiogenesis, and several studies have demonstrated that TEX mimic the proangiogenic properties of their cells of origin (10, 12, 15). TEX deriving from PCI-13 and UMSCC47 mainly carried uPA, which could be a putative stimulator in the functional assays with HUVECs in our study. The uPA/uPAR system is an important pathway for activating pericellular proteolysis, increasing vascular permeability and by supporting EC proliferation and migration and, therefore, stimulating angiogenesis (22). In addition to the transfer of proteins to target cells, genetic reprogramming of ECs was demonstrated in several studies. This suggests that the transfer of mRNAs and miRNAs are essential underlying mechanisms used by TEX for reprogramming of recipient cells, including ECs (10, 23).

Exosomes can use a variety of internalization mechanisms for entering recipient cells, such as endocytosis (clathrin or caveolin dependent), phagocytosis, micropinocytosis, or lipid raft–mediated internalization (24). We used a spectrum of pharmacologic inhibitors to block specific pathways of vesicle uptake to investigate, which pathway is mainly involved in the exosome uptake by ECs. The cholesterol depletion drug MβCD was used to block lipid raft/caveolin-dependent endocytosis, as well as the inhibitors genistein and nystatin, which disrupt lipid rafts and inhibit the tyrosine protein kinase (19). All those inhibitors blocked the exosome internalization at least in part, suggesting, that the lipid raft/caveolin–dependent endocytosis is one pathway of exosome internalization by ECs. Another main route of vesicle uptake is the receptor-mediated endocytosis. Our data implicate this pathway, because a heparin treatment led nearly to a full inhibition of exosome uptake. Heparin is a competitive inhibitor of cell-surface receptors dependent on heparan sulfate proteoglycan (HSPG) coreceptors (25, 26). Our data suggest that TEX can use different internalization mechanisms, similar to other cell types including dendritic cells, macrophages, or cancer cells (24).

Our data reveal that TEX are potent stimulators of angiogenesis in an orthotopic HNSCC mouse model. The mechanisms underlying our data remain still unclear. One main effect of TEX is the modulation of the TME and the formation of a premetastatic niche (27–29). Therefore, the preparation of the TME by TEX could pave the way for the initial tumorigenesis and the subsequent tumor progression, including accelerated angiogenesis. Interestingly, we observed not only increased endothelial structures in the 4-NQO tumors, but also increased numbers of pericytes, which is likewise reported for exosome-treated glioma cell–induced xenografts (15). Pericytes have diverse functions including the guidance of sprouting tubes, the sense of angiogenic stimuli and mainly the stabilization of vessels (30). They reflect the status of a mature vessel and are therefore heavily involved in angiogenic processes. Future studies will be necessary to evaluate the interactions of TEX and pericytes, especially because only a few studies focused on this important component of tumor angiogenesis so far (15, 31).

In summary, TEX appear to play an important role in tumor angiogenesis and thus may contribute to tumor growth and metastasis in this context. To further improve existing therapeutic modalities and long-term therapeutic results for patients with HNSCC, a detailed understanding of the pathogenesis of HNSCC, including TEX, is necessary. Because HNSCCs are characterized by strong vascularization, it would be a significant advancement to identify mechanisms used by TEX to promote angiogenesis and to acquire an understanding of how to effectively inhibit exosome-driven angiogenesis.

No potential conflicts of interest were disclosed.

Conception and design: N. Ludwig, T.L. Whiteside

Development of methodology: N. Ludwig, S.S. Yerneni

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Ludwig, S.S. Yerneni, B.M. Razzo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Ludwig, S.S. Yerneni

Writing, review, and/or revision of the manuscript: N. Ludwig, T.L. Whiteside

Study supervision: T.L. Whiteside

The authors are grateful to Drs. Chang-Sook Hong and Priyanka Sharma for their advice and technical assistance. This work was supported in part by NIH grants R0-1 CA168628 and R-21 CA205644 (to T.L. Whiteside). S.S. Yerneni was supported by the Dowd Fellowship at Carnegie Mellon University. In vivo study was supported by the NIH grant R0-1 CA168628 05S1 (to T.L. Whiteside and B.M. Razzo). N. Ludwig was supported by the Leopoldina Fellowship LPDS 2017-12 from German National Academy of Sciences Leopoldina.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.