Patients (n=18) from the department of otorhinolaryngology of the University hospital of Basel (Switzerland) with HNSCC in early and advanced stages were included as well as 3 healthy controls (Table I). The categorization is based on the current TNM 8 (16). The classification is used to divide malignant tumors into stages. The three main categories of the TNM system correspond to the three letters: T=tumor, extent and behavior of the primary tumor; N=nodus-absence or presence of regional lymph node metastases; M=metastases, absence or presence of distant metastases. Depending on this, a distinction is made between low grade and high grade (WHO I + II vs. III + IV). Ethical approval (approval no. 2020-02173) was issued by ethical commission of the northwest and central Switzerland. All patients provided written informed consent for research and consented to anonymous processing of the blood samples collected for scientific purposes. Blood collection was performed before the start of tumor therapy. A standardized peripheral venous blood collection of 7.5 ml blood was performed (EDTA; S-Monovette; Sarstedt). To remove cells and cell debris, whole blood was further centrifuged at 800 g, 10 min at room temperature. Platelet-rich plasma (PRP) was collected into new tubes. Platelet-poor plasma (PPP) was generated from PRP by additional 10,000 g centrifugation at 4°C for 30 min, aliquoted and stored at −80°C (7,12).

The EXÖBead^® used are magnetic beads coated with galectins for the isolation of EVs. The basis of this was the detection of N-linked glycoproteins on EVs, in particular galectin-3-binding protein (LGALS3BP), which was found on EVs from ovarian cancer cells (17-19).

PPP (1 ml) was diluted directly in 0.9 ml of 0.5% EV-free BSA (SERVA Electrophoresis GmbH) in PBS (100,000 g, 4°C for overnight centrifugation) and incubated with 100 µl of EXÖBead^® (1 µm, 6×10⁸ particles/ml, Biovesicle Inc.) for 1 h at 25°C. EV-EXÖBead^® complexes were washed twice with 1 ml 0.5% EV-free BSA in PBS while magnetic EXÖBead^® were kept in the tube by using a magnet (Biovesicle, Inc.). A total of 1 ml of 10% EV-free BSA in PBS was also incubated with EXÖBead^® as a negative control (EF-EXÖBead^® complex). Unbound plasma was also collected as the non-EVs fraction. Non-EVs fractions were further used in intracellular EV marker and non-EVs marker staining. EVs-EXÖBead^® complexes and EF-EXÖBead^® complex were further analyzed regarding EV surface marker, intracellular EV marker and non-EV marker staining. To dissolve EVs from the beads, EVs-EXÖBead^® complexes were incubated with 200 µl of 0.3 M lactose in PBS with EV-EXÖBead^® complexes at room temperature for 1 h. EF-EXÖBead^® complex was also incubated with 0.3 M lactose in PBS as an elution buffer negative control. Particles from EF-EXÖBead^® complex were too low to detect (data not shown). Eluted EVs were further analyzed by nanoparticle Tracking Analysis (NTA), transmission electron microscopy (TEM), cryogenic TEM (cryo-TEM) and peripheral blood mononuclear cells (PBMCs) functional assay (Fig. 1).

Antibody master mix (200 µl) was incubated each with plasma EV-EXÖBead^® complexes as well as with EV-free (EF)-EXÖBead^® complexes (1 ml of 10% EV-free BSA in PBS). Incubation lasted for 1 h at 25°C. After antibody incubation, the EV-EXÖBead^® and EF-EXÖBead^® complexes were washed twice with 1 ml of 0.5% EV-free BSA in PBS, while the magnetic EXÖBead^® were held in the tube using a magnet (Biovesicle, Inc.). Plasma EV-EXÖBead^® and EF-EXÖBead^® complexes stained with antibodies were subsequently prepared on the BD LSR Fortessa™ (BD Biosciences) and the data were analyzed using FlowJo software (Tree Star, Inc.). EF-EXÖBead^® complexes served as a non-EV control. Plasma EV-EXÖBead^® complexes were also stained with IgG antibodies. The background signal of EF-EXÖBead^® complexes with test master mix antibodies and plasma EVs-EXÖBead^® complexes with IgG antibodies was comparable (data not shown). EF-EXÖBead^® complexes were used as the gating basis of the flow cytometry data. Our antibody master mix contained: 2.5 µg/ml of PE/Cyanine7 anti-human CD63 (1:80; clone: H5C6), FITC anti-human CD81 (1:80; clone: 5A6), APC anti-human CD9 (1:80; clone: HI9a; all from Biolegend, Inc.) and eFluor 450 anti-human programmed death-ligand 1 (PD-L1; 1:80; clone: MIH1; Thermo Fisher Scientific, Inc.). For the EV subpopulation gating strategy, CD9^{+ or Neg} and CD81^{+or Neg} were first gated on Plasma EV-EXÖBead^® complexes which based on the background signal from EF-EXÖBead^® complexes. These four-population included CD9⁺ CD81^Neg, CD9⁺ CD81⁺, CD9^Neg CD81⁺ and CD9^Neg CD81^Neg plasma EV-EXÖBead^® complexes. These four populations were further gated with CD63^{+ or Neg} and PD-L1^{+ or Neg} based on the signal of EF-EXÖBead^® complexes. Significance was calculated by two-way ANOVA with Šídák's multiple comparisons test.

Due to limited fluorescence channels, the PanEV antibody was used as a positive EV marker instead of three individual antibodies. PanEV antibody (3 µl) was premixed with 1 µl of PE anti-CD63: H5C6 clone, 1 µl of PE-CD81: 5A6 clone and 1 µl of PE-CD9: HI9a clone (1:66.7; Biolegend, Inc.). The antibody master mix contained the PanEV marker, three disease-specific biomarkers, and a common leukocyte antigen CD45. The mix was set up of 3 µl of PE-PanEV antibodies (1:66.7), 4 µl of Alexa Fluor^® 488 anti-Pan Cytokeratin: C-11 (10 µg/ml; 1:50), 5 µl of Brilliant Violet 421™ anti-human CD45 Antibody: HI30 (1:40; all from Biolegend, Inc.), 5 µl of APC anti-PD-L1 antibody: MIH1 (1:40; Thermo Fisher Scientific, Inc.) and 5 µl of APC-Cy7 Mouse Anti-Human CD326 antibody: 9C4 (APC-Cy7 Mouse Anti-Human CD326 antibody; 1:40). To avoid a multiple plasma EV-EXÖBead^® aggregation signal, individual beads from FSC-A and SSC-A were first gated. For the single positive EV subpopulation gating strategy, 'PanEV⁺ and CD45^{+ or Neg}', 'PanEV⁺ and epithelial cell adhesion molecule (EpCAM)^{+ or Neg}', 'PanEV⁺ and PD-L1^{+ or Neg}' or 'PanEV⁺ and pan-cytokeratin (PanCK)^{+ or Neg}' were further gated on Plasma EV-EXÖBead^® complexes which based on the background signal from EF-EXÖBead^® complexes. The final populations were expressed as a percentage of single beads. For the double positive EV subpopulation gating strategy, PanEV⁺ and CD45^Neg were gated on plasma EV-EXÖBead^® complexes based on the background signal from EF-EXÖBead^® complexes. PanEV⁺ CD45^Neg plasma EV-EXÖBead^® complexes were further gated EpCAM^+/Neg and PD-L1^+/Neg also based on the background signal from EF-EXÖBead^® complexes. The final four double positive populations of PanEV⁺ CD45^Neg EpCAM⁺ PD-L1^Neg, PanEV⁺ CD45N^eg EpCAM⁺ PD-L1⁺, PanEV⁺ CD45^Neg EpCAM^Neg PD-L1⁺ and PanEV⁺ CD45^Neg EpCAM^Neg PD-L1^Neg were expressed as percentages of single beads. For the triple positive EV subpopulation gating strategy, PanEV⁺ and CD45^+/neg were gated on Plasma EV-EXÖBead^® complexes based on the background signal from EF-EXÖBead^® complexes. Two populations of PanEV⁺ CD45^Neg and PanEV⁺ CD45⁺ EV-EXÖBead^® complexes were further gated with EpCAM^+/Neg and PanCK^+/Neg also based on background signal from EF-EXÖBead^® complexes. These four populations of PanEV⁺ CD45^Neg with EpCAM^+/Neg and PD-L1^+/Neg were further gated by PD-L1⁺ population based on background signal from EF-EXÖBead^® complexes. The final four triple positive populations of PanEV⁺ CD45^Neg EpCAM⁺ PanCK^Neg PD-L1⁺, PanEV⁺ CD45^Neg EpCAM⁺ PanCK⁺ PD-L1⁺, PanEV⁺ CD45^Neg EpCAM^Neg PanCK⁺ PD-L1⁺ and PanEV⁺ CD45^Neg EpCAM^Neg PanCK^Neg PD-L1⁺ were expressed as percentages of single beads. Another final four triple positive population of PanEV⁺ CD45⁺ EpCAM⁺ PanCK^Neg PD-L1⁺, PanEV⁺ CD45⁺ EpCAM⁺ PanCK⁺ PD-L1⁺, PanEV⁺ CD45⁺ EpCAM^Neg PanCK⁺ PD-L1⁺ and PanEV⁺ CD45⁺ EpCAM^Neg PanCK^Neg PD-L1⁺ also used the same gating strategy and were expressed as percentages of single beads. Significance was calculated by an unpaired t-test with Welch's correction. The specific test was selected based on number of groups, number of samples, and normality.

Non-EV fractions were collected after PPP incubation with EXÖBead^®. To compare the non-EVs fraction to the plasma EV-EXÖBead^® complexes, the non-EVs fraction were coupled to the same amount, number, and material of magnetic beads. The coupling steps consisted of first activating the magnetic particles with carboxyl groups (Chemicell GmbH) through incubation with 1-Ethyl-3-[3-dimethylaminopropyl] carbodiimide hydrochloride (EDC; Merck KGaA) in 0.1 M 2-(N-Morpholino) ethane sulfonic acid (MES; cat. no. M8250; Merck KGaA) at pH 5.0 for 30 min at 25°C. Then non-EV fractions were covalent-coupled to the same number and same material of EDC activated carboxyl-group magnetic particles by incubation for 2 h at 25°C. Non-EV-bead complex were further blocked by 0.5% EV-free BSA in PBS. A total of 1 ml of 10% EV-free BSA in PBS was coupled with EDC-activated carboxyl-group magnetic particles as a negative control. Plasma EV-EXÖBead^® complexes, EF-EXÖBead^® complexes, non-EV fractions-bead complexes and EF-bead complexes were then fixed by 200 µl of 4% paraformaldehyde in PBS for 10 min at 25°C, penetrated by 200 µl of 0.1% Tween-20 in PBS for 10 min at 25°C and blocked by 200 µl of 10% EV-free FBS in PBS (100,000 g, 4°C for overnight centrifugation) for 30 min at 25°C. Fixed/Penetrated plasma EV-EXÖBead^® complexes, EF-EXÖBead^® complexes, non-EV fractions-bead complexes and EF-bead complexes were further stained with 200 µl of 0.5% EV-free BSA in PBS with antibodies master mix at 25°C for 1 h incubation. Antibodies master mix contained 2.5 µg/ml of Alexa Fluor^® 488 anti-apolipoprotein A1/ApoA1 (1:40; clone: 2083A; R&D Systems, Inc.) and 5 µg/ml of PE anti-TSG101 (1:40; clone: EPR7130(B), Abcam). Antibodies-stained plasma EV-EXÖBead^® complexes, EF-EXÖBead^® complexes, non-EV fractions-bead complexes and EF-bead complexes were visualized by BD LSRFortessa™ (BD Biosciences) and data was analyzed with FlowJo V10.8 software (Tree Star, Inc.). The geometric mean fluorescence intensity (MFI) was used for the calculation. The initial MFIs of EF-EXÖBead^® complexes and EF bead complexes were similar (data not shown). For the group of plasma EV-EXÖBead^® complexes, the reduced MFI of the negative control was the MFI of the plasma EV-EXÖBead^® complexes minus the MFI of the EF-EXÖBead^® complexes. For the non-EV fraction bead complexes, the reduced MFI of the negative control was the MFI of the non-EV fraction bead complexes minus the EF bead complexes. Significance was calculated using an unpaired t-test.

Size-exclusion chromatography (SEC) was used qEV original 70 nm (IZON Science, Ltd.). EV fractions were collected from fraction 7 to 10 as recommends by the manual. A total of 500 µl of solution was collected per fraction. A total of 7 to 10 fractions were then pooled together as SEC-EVs (2 ml). To understand the total isolated components of SEC isolation, SEC was not concentrated by further steps such as ultracentrifugation or ultrafiltration. SEC EVs (2 ml) were coupled with the same amount, number and material of magnetic beads. SEC EVs were further coupled to 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) activated carboxyl group magnetic particles (Chemicell GmbH). A total of 1 ml of 10% EV-free BSA in PBS was also coupled with EDC-activated carboxyl group magnetic particles as a negative control. In the comparative study, plasma EV-EXÖBead^® complexes, EF-EXÖBead^® complexes, non-EV fraction bead complexes, and EF bead complexes were incubated with an antibody master mix in 200 µl of 0.5% EV-free BSA for 1 h at 25°C. The antibody master mix contained 3 µl PE-PanEV (premixed CD9, CD63 and CD81 antibodies; 1:66) and 5 µl apolipoprotein A-I/ApoA1 Alexa Fluor^® 488-conjugated antibodies (ApoA1; 1:40; cat. no. EP1368Y; Abcam). The initial MFIs of EF-EXÖBead^® complexes and EF bead complexes were similar (data not shown). Plasma EV-EXÖBead^® complexes, EF-EXÖBead^® complexes, EF-bead complexes, and SEC-EV-bead complexes were visualized on BD LSRFortessa™ (BD Biosciences) and the data were analyzed using FlowJo software (Tree Star, Inc.). For the gating strategy, a gating population of beads was first determined based on FSC-A and SSC-A. Then, the PanEV^+/Neg and ApoA1^+/Neg populations were gated based on EF control. Significance was calculated using an unpaired t-test.

EV numbers and sizes were determined by ZetaView^® Nanoparticle Tracking Analyzer PMX 110 (Particle Metrix GmbH). EVs were diluted in PBS to a final volume of 1 ml. For each measurement, two measurement cycles were performed by scanning 11 positions each and acquiring 60 frames per second under the following settings: pre-acquisition parameters were set to a sensitivity of 80; Shutter was set to 70. Cell temperature was set to 25°C and trace length to 15 (20). After recording, the videos were analyzed with the built-in ZetaView software 8.05.11 SP1 (Particle Metrix GmbH) with specific analysis parameters: minimum particle brightness: 20; minimum size of 5 pixels, maximum size of 1,000 pixels and PSD nm/class of 10 and PSD classes/decade of 10.

A total of 5 µl of the undiluted sample were adsorbed for 60 sec to glow-discharged parlodion/carbon-coated copper grids. The grids were then blotted, washed 3 times with double-distilled water and negatively stained on two droplets of 2% uranyl acetate solution. Samples were imaged using a FEI Talos F200C TEM (FEI) operated at 120 kV. Electron micrographs were recorded on a Veleta Camera (EMSIS GmbH).

A total of 4 µl aliquot of sample was adsorbed onto holey carbon-coated grid (Ted Pella, Inc.) blotted with Whatman 1 filter paper and vitrified into liquid ethane at -178°C using a Leica GP plunger (Leica Microsystems GmbH). Frozen grids were transferred onto a Talos electron microscope (FEI) using a Gatan 626 cryo-holder. Electron micrographs were recorded at an accelerating voltage of 200 kV, using a low-dose system (20 e-/Å2) and keeping the sample at low temperature. Micrographs were recorded on a CETA camera (Thermo Fisher Scientific, Inc.).

The functional assay was performed in CD4⁺ immune cells of 3 patient blood sample in 3 replicates. CD4⁺ cell extraction was performed from buffy coats obtained from the Blood Donation Center (Basel District) according to the manufacturer's protocol using the StraightFromTM Buffy Coat CD4 Micro Bead kit (product no; 130-114-980 MACS Miltenyi Biotec, Inc.). The T-cell activation was conducted according to the manual as described in the T Cell Activation/Expansion kit (product no. 130-091-441; MACS Miltenyi Biotec, Inc.) (21,22). A total of 30 out of 200 µl eluted EVs isolated from tumor patient plasma were added to each well and gently mixed for 24 h at 37°C. Cells were further stained with an antibody master mix containing 2.5 µl anti-body/test of CD152 Monoclonal Antibody APC (Clone: 14D3), CD4 Monoclonal Antibody eFluor^® 450 (Clone: SK3) and CD69 Monoclonal Antibody FITC: FN50 (all from Thermo Fisher Scientific, Inc.) for 50 min at 37°C. Analytical assay was performed using flow cytometry (CytoFlex; Beckman Coulter, Inc.). Significance was calculated by non-parametric Kruskal-Wallis test with Dunn's multiple comparisons test.

The functional assay was performed in PBMCs of different HNSCC patients' plasma EVs in technological triplicates. PBMCs were isolated from buffy coats received by healthy donors. PBMCs activation was conducted according to the manual as described in the T Cell Activation/Expansion kit (product no. 130-091-441; MACS Miltenyi Biotec, Inc.). Eluted EVs (5×10⁷) isolated from tumor patient plasma were added to each well with 1×10⁶ PBMCs/ml in 1 ml culture medium and gently mixed for 24 h at 37°C in the incubator. Lactose elution buffer (0.3 M lactose in PBS) was used as a negative control and as a gating base. To avoid the effect of the elution buffer, the total amount of CD69 was also examined in the elution buffer treatment group and in the group with only activated T cells. The total CD69 amount did not differ between elution buffer with T-cell activation stimulation and T-cell activation stimulation only (data not shown). Cells were first stained at 100 µl at a 1:100 ratio with the cell viability dye Zombie NIR™ for 30 min at 25°C (Biolegend, Inc.). After two washing steps, cells were stained with 100 µl of 1:100 antibody master mix, CD4 Monoclonal Antibody eFluor^® 450 (Clone: SK3, Thermo Fisher Scientific, Inc.), APC anti-PD-L1 antibody: MIH1 (Thermo Fisher Scientific, Inc.), PE anti-human CD279 (PD-1) Antibody: EH12.2H7 (Biolegend Inc.) and CD69 Monoclonal Antibody FITC: FN50 (Thermo Fisher Scientific, Inc.) for 30 min at 4°C. Analytical assay was performed using flow cytometry: BD LSRFortessa™ (BD Biosciences) and flow data were analyzed by FlowJo software (Tree Star, Inc.). First, the lymphocyte population was recorded based on FSC-A and SSC-A. Then, the population of individual cells was detected based on FSC-A and FSC-H. To identify live cells, the zombie NIR negative population was gated. To examine live T cells, the CD4⁺ population was then gated. The final gate was gated to CD69^+/Neg with PD1^+/Neg or CD69^+/Neg with PD-L1^+/Neg. Significance was calculated by Brown-Forsythe and Welch's ANOVA test with Dunnett's T3 multiple comparisons test.

Flow cytometry data including gating were performed by FlowJo version 10.8 software (Tree Star, Inc.). Data were statistically analyzed with Prism version 8 (GraphPad Software, Inc.). The specific test used is described in the figure legend. P<0.05 was considered to indicate a statistically significant difference.

To compare whether the eluted EVs were similar in size to those reported in the literature (23) the morphology was examined after isolation of EXÖBead^® and elution in lactose-PBS buffer in conventional TEM, where they were observed as intact, cup-shaped, membrane-bound vesicles with a size of 30-200 nm (Fig. 2A and B). Eluted EVs were observed as a double lipid layer of vesicles in cryo-EM with a size of 30 nm-250 nm (Fig. 2C and D). ZetaView (Particle Metrix, Inning am), a NTA instrument, was used to measure the yield and size of nanoparticles from three individual donors as biological triplicates (Fig. 2E-G) and three technological triplicates of a same donor (Fig. 2H). The medium size of eluted EVs from three technological triplicates was 81.6±2.43 nm (CV: 2.97%; Fig. 2E-G). The total particle number from 1 ml plasma of 3 technological triplicates was 1.57×10⁸±2.8×10⁷ (CV: 13.29%; Fig. 2H). The eluted EVs from EXÖBead^® isolation represented reproducible and high similarity as mentioned in the literature (23).

The recovered EVs were analyzed and checked for established EV surface markers by bead-based flow cytometry. CD63⁺, CD9⁺ and CD81⁺ were reported as common and MISEV compliant EV surface markers (23). Geometric MFI of CD63, CD9 and CD81 was expressed on recovered EXÖBead^® isolates and was increased in HNSCC patients compared with healthy donors for all surface markers (Figs. 3A and S1). MFI of CD9⁺ was significantly higher in patients with HNSCC compared with the healthy controls (P=0.0097; Figs. 3A and S1A). In addition, MFI of PD-L1⁺, an important diagnostic cancer and surface marker (24) was increased in HNSCC tumor patients compared with the healthy controls. Relative ratio of PD-L1⁺ MFI versus CD81⁺ MFI (P=0.0004) was significantly higher in patients with HNSCC compared with the healthy controls (Fig. 3B) in both cases. PD-L1⁺ CD9⁺ CD63⁺ CD81^Neg EVs-EXÖBead^® complex from HNSCC patients had also higher relative scores (P<0.0001; Figs. 3C and S1B-E). However, PD-L1⁺ CD9⁺ CD63⁺ CD81⁺, PD-L1⁺ CD9⁺ CD63^Neg CD81⁺ and PD-L1⁺ CD9⁺ CD63^Neg CD81^Neg EVs-EXÖBead^® complex showed no difference between the two groups (Figs. 3C and S1B-E). These population changes suggested that PD-L1⁺ CD9⁺ CD63⁺ EVs may play an important role in HNSCC, but further research is required to support or refute this assumption. In addition to extracellular markers, TSG101 was additionally measured as typical intracellular EV marker (23). The experiments were well feasible and TSG101 could be detected in EVs of HNSCC after isolation and recovery by EXÖBead^®. The levels of the isolates were significantly increased in the EXÖBead^® fraction compared with the unbound patient plasma (P=0.015; Figs. 4A and S2A). A frequent problem of EV isolation is the contamination by protein aggregates or other aggregates/particles from blood samples, due to the complexity of blood composition (23). A well-established marker for contamination in EV preparations is ApoA1 (23). The ApoA1 concentration was high in residual plasma with MFI 7000 compared with almost none in the EXÖBead^® fraction (P=0.0002; Figs. 4A and S2A). To gain an improved understanding of the specificity of EV isolation, SEC was compared with EXÖBead^® in terms of PanEV and ApoA1 protein presence. PanEV⁺ ApoA1^neg population was significantly higher in the EV-EXÖBead^® complex compared with the EV-SEC beads complex (P=0.019; Figs. 4B and S2B). While PanEV^neg ApoA1⁺ population was significantly higher in EV-SEC beads complex compared with EV-EXÖBead^® complex (P=0.0061; Figs. 4B and S2B). This flow cytometric result of EV-EXÖBead^® complex suggested that measurable lipoprotein contamination may be lower with EXÖBead^® isolation of plasma EVs than with isolation by SEC method. Notably, PanEV⁺ ApoA1⁺ population may be detected by both isolation methods with no significant difference (23,25).

A specific set of serum markers for diagnosis and therapeutic purpose in HNSCC patients is currently lacking (26). CD45, EpCAM, PanCK and PD-L1 have been reported as potential biomarkers in HNSCC patients and their isolated EVs/exosomes (11,24). However, the expression level of CD45, EpCAM, PanCK and PD-L1 on EVs in HNSCC patients remains unknown. EXÖBead^® was used to isolate EVs from plasma which were then stained with the master mix of antibodies containing PanEV, CD45, EpCAM, PanCK and PD-L1. The expression level of EpCAM⁺ in HNSCC patients was significantly higher than in healthy controls (P=0.023) while there was no significant difference in the other markers when considered independently (Figs. 5A and S3A). To accurately determine the percentage of tumor-specific EVs in the circulation, single EV bead complexes were next gated to exclude multiple EV bead complexes and examine the difference between HNSCC patients and healthy controls. In the antibody staining of the obtained EVs, the results of EV-EXÖBead^® complex showed that there was a significant difference between CD45^neg PanEV⁺, EpCam⁺ PanEV⁺ and PDL1⁺ PanEV⁺ between patients with HNSCC and healthy controls (CD45^neg PanEV⁺, P=0.026; EpCAM⁺ PanEV⁺, P=0.004; PD-L1⁺ PanEV⁺, P=0.007) (Figs. 5B-D and S3B-D). The EV-EXÖBead^® complex stained with PanCK showed no significant difference between HNSCC and control (Figs. 5E and S3E). The EV-EXÖBead^® complex population of CD45^neg EVs that was EpCAM⁺ and PD-L1⁺ was significantly increased in HNSCC patients compared with the control group (P=0.002; Fig. 5F and Fig. S3F-I). The same was found for the group of EV-EXÖBead^® complex which was CD45^neg, EpCAM⁺ and PD-L1^neg (P=0.003; Figs. 5F and S4F-I). For this purpose, CD45^neg, EpCAM^+/neg, PD-L1^+/neg and PanEV⁺ were gated on a single EV beads complex (Fig. 5F and S3F-I). In addition, the triple positive population on the EV-EXÖBead^® complex was examined. It was found that CD45^neg EV-EXÖBead^® complexes that were double or triple positive for the markers EpCAM, PD-L1 and PanCK were significantly more expressed in HNSCC than in healthy controls [(EpCAM⁺, PD-L1⁺ and PanCK^neg; P=0.0005) (EpCAM⁺, PD-L1⁺ and PanCK⁺; P=0.02; Figs. 5G and S3J-M)]. Notably, it was also identified that CD45⁺ EV-EXÖBead^® complexes with EpCAM⁺, PD-L1⁺, but PanCK⁻ were detectable to a significantly higher extent in HNSCCs than in healthy controls (P=0.025; Figs. 5H and S3N-Q). These significantly altered populations of EV-EXÖBead^® complex suggested that single positive EVs, double positive EVs or triple positive EVs may play an important role in the progression of HNSCC. Further studies need to be performed. To demonstrate yet again that EVs were specifically isolated with the method described in the present study, particle size and number of particles were checked in parallel with the analysis of specific markers on the EVs with NTA. Similar particle sizes were found in both groups (HNSCC: 125.2±9.1 nm; control: 123.32±7.42 nm). The particle number was comparable (HNSCC: 2.54×10⁸±1.27×10⁸; control group: 5.58×10⁸±3.17×10⁸) (Fig. 5I).

To evaluate the biological activity and functional in vitro competence of EVs isolated and recovered from plasma of tumor patients, EVs obtained by EXÖBead^® as well as EVs obtained by polyethylene glycol (PEG)-based precipitation method was co-incubated with activated T cells (CD4⁺ T cells) and compared with activated T cells without EV co-incubation. The aim was to measure the in vitro immunosuppressive capacity of EVs isolated by the established PEG-based precipitation method (27,28) and compare it to the immunosuppressive capacity of EVs isolated by EXÖBead^®. In the present study, the purity of CD4⁺ T cells was at least 90%, analogous to analyses in older proprietary trials (6). Percentage of CTLA4⁺ CD69^neg T responder cells was measured by flow cytometry (29,30) after co-incubation with EVs isolated by the two different methods. A highly significant difference was observed between untreated activated T cells and T cells treated with 30 out of 200 µl EXÖBead^®-isolated EVs in terms of CTLA4 and CD69 expression (P=0.025; Fig. 6A and S4A). Notably, this effect was not observed when activated T cells were treated with EVs obtained by PEG-based isolation (Figs. 6A and S4A). There was no difference of single CTLA4⁺ CD4⁺ T cells within these three groups (Fig. S4B). To address the question of whether EVs isolated from plasma of patients with HNSCC using EXÖBead^® retain their biological activity, EVs from different patients were examined for immunosuppression. The present data showed that PD-L1⁺ CD69^neg CD4⁺ T cells were significantly increased after co-incubation with 5×10⁷ plasma EVs from HNSCC patients (n=13, with technological triplicate) compared with co-incubation with elution buffer (technological triplicate) (P=0.0011) and healthy donors (n=3, with technological triplicate) (P=0.0045) as controls (Figs. 6B and S4C). It was also found that PD-L1⁺ CD69^neg CD4⁺ T cells were also higher in treatment with healthy controls' EVs compared with controls (P=0.0404; Figs. 6B and S4C). Notably, it was also found that PD1⁺ CD69^neg CD4⁺ T cells were also significantly increased in patients' group compared with the elution buffer group (P=0.0344), but there was no difference between healthy controls and elution buffer group (Fig. S4D). No difference was observed between the groups of patients and controls and elution buffer from signal positive of CD69, PD1 and PDL1 CD4⁺ T cells (Fig. S4E). Thus, by isolating EVs with EXÖBead^® it appeared to be possible to isolate functionally competent EVs from plasma of HNSCC patients.