Differential expression analysis of HNSCC patient transcriptomes, taken from The Cancer Genome Atlas (TCGA), was performed using the R Studio software with packages ‘BiocManager’, ‘TCGAbiolinks’, ‘limma’ and ‘edgeR’ (20–24). A glmLRT fit was performed with FDR and logFC cut-offs set to 0.01 and 1, respectively. Kaplan-Meyer plot of patient survival analysis was generated using the Xena Functional Genomics Explorer (University of California, Santa Cruz) with default settings including a statistical analysis with log-rank test (25).

HPV-negative HNSCC cell lines BHY (DSMZ), FaDu (ATCC), CAL-33 (DSMZ) and SCC131 (DSMZ) were cultivated in a humidified atmosphere at 37°C and 5% CO₂. DMEM (Dulbecco’s modified Eagle’s medium, Gibco) with GlutaMAX, pyruvate and high Glucose was used for BHY and CAL-33 cells. The same medium with low Glucose was used for FaDu cells. Minimal Essential Medium (MEM, Gibco) with GlutaMAX and Earle’s Salts was used for SCC131. The media were supplemented with 10% FBS (fetal bovine serum, GE Healthcare). Green Fluorescent Protein (GFP)-expressing derivatives of the cell lines BHY and FaDu (BHY-GFP and FaDu-GFP), generated as described previously (17), were cultivated in their respective media with additional 0.3 µg/ml or 0.1 µg/ml puromycin to maintain stable GFP expression.

Cell line identity was confirmed by genomic sequencing of nine marker loci: D5S818, D13S317, D7S820, D16S539, VWA, TH01, AM, TPOX, CSF1PO (Eurofins Genomics). Mycoplasma negative status was confirmed with MycoAlert (Lonza).

For GRP78 overexpression, 2.5 x 10⁵ cells were transfected with pcDNA3.1(+)-GRP78/BiP (32701, Addgene) or empty control vector (V79020, Invitrogen) using Lipofectamine 2000 Transfection Reagent (Invitrogen) according to the manufacturers protocol.

The GRP78 inhibitor HA15 (Selleck Chemicals) was diluted in DMSO with pure DMSO serving as control. FaDu and FaDu-GFP cells were treated with 1.5, 2.5 or 3.5 µM, BHY and BHY-GFP cells were treated with 20, 30 or 40 µM inhibitor.

Bovine EVs were removed from FBS by ultracentrifugation (100,000g, 4°C, 14h) to generate EV-depleted FBS (EdFBS).

Cells were X-irradiated at room temperature with a dose rate of 0.82 Gy/min (Xstrahl RS225 X-ray system at 195 kV and 10 mA with a 3 mm aluminum filter). Sham irradiated cells were treated identically, without exposure.

Cells were detached with 0.05% trypsin or collected with a cell scraper to retain surface proteins. Cell pellets were lysed at 4°C with T-PER lysis buffer (Thermo Fisher) combined with PhosSTOP and cOmplete, phosphatase and protease inhibitor cocktails (Roche). Protein concentrations were determined with the Pierce BCA Protein Assay Kit (Thermo Fisher).

For immunoblotting, either 10 µg protein or 15 µl of EV suspension were used. Primary antibodies used were Alix (2171, Cell Signaling), Beta-actin (SAB1305567, Sigma-Aldrich), Calnexin (sc11397, Santa Cruz), GAPDH (sc-47724, Santa Cruz), GRP78/BiP (3177, Cell Signaling), TSG101 (GTX70255, GeneTex) and CD9 (sc-13118, Santa Cruz). Proteins were revealed using horseradish-peroxidase conjugated secondary antibodies (anti-mouse: sc2005, anti-rabbit: sc2005; Santa Cruz) and the Amersham ECL Select Western Blotting detection reagent (GE Healthcare). Digital images were captured with the FluorChem HD2 (Alpha Innotec).

3 x 10⁵ seeded cells were harvested 24 h after irradiation with 6 and 0 Gy. After detaching with a cell scraper for intact surface proteins, cells were incubated with anti-GRP78 antibody followed by a secondary Alexa Fluor 488-coupled antibody (anti-rabbit: A-11008, Thermo Fisher). Incubation steps were alternated with PBS+0.5% bovine serum albumin washing steps. A minimum of 10,000 labelled cells were analyzed with a FACSCAN LSRII (Becton-Dickinson). To remove cellular fragments and dead cells from the analysis, cells incubated with secondary antibody only were used for gating according to their forward and sideward scatter properties. Afterwards, doublets were excluded by plotting FSC height vs. FSC area, remaining particles were analyzed in terms of fluorescence intensity and their mean intensity was documented and used for calculations.

For analysis of GRP78 expression on the EV-surface, 10 µl EVs (1 x 10⁹ EV/ml) derived from 0 or 6 Gy-irradiated BHY cells were absorbed onto 10 µl latex beads (A37304, Invitrogen) for 1 h and in 100 µl PBS overnight. Following 30 min incubation with PBS+1 M Glycine, cells were labelled with either anti-CD63 (sc15363, Santa Cruz) or anti-GRP78/BiP antibodies followed by Alexa Fluor 488-coupled secondary antibody. All incubation steps were performed at 4°C.

Transfected FaDu and BHY cells (see section Cell Culture) were incubated for 48 h and seeded in 6-well plates at a range of cell densities. 24 h later, they were irradiated with different radiation doses ranging from 0 to 8 Gy and colonies were expanded for 10-14 days. Ethanol-fixed colonies were stained with Giemsa (1:10 in PBS, Boehringer Ingelheim) and counted. Survival fractions after 2 Gy-irradiation (SF2) were determined with R Studio.

Gap closure assays were performed using GFP-expressing cells following a previous protocol (17). 55,000 BHY or 75,000 FaDu cells were seeded into the individual wells of removable 12-well silicone grids (Ibidi), that were placed in 10-cm dishes. After 24 h, the silicone spacers were removed to create even gaps between the individual cell monolayers and 8 ml medium was added.

For inhibition of GRP78, the medium, added after removal of the silicone grids, contained HA15 or DMSO as control.

For overexpression of GRP78, the cells were seeded as stated above 48 h after transfection and normal medium was used.

Time-lapse images of the GFP-labelled cells were captured by fluorescence microscopy with a Biorevo BZ-9000 (Keyence). Migratory behavior was quantified from these images with the software package ‘countcolors’ in R Studio (26). A pixel intensity cut-off for green pixels of 20 (range: 0-255, RGB color space) was chosen based on the green pixel intensity of the 0h-background.

1.25 x 10⁶ BHY cells, 1.5 x 10⁶ FaDu cells and 2 x 10⁶ CAL-33 or SCC131 cells were seeded with 8 ml medium in 10-cm dishes and irradiated with 0 and 6 Gy after 48 h. Prior to irradiation, the cells were washed with PBS and EdFBS-supplemented medium was added. 24 h after irradiation, small EVs were collected from the supernatant by serial ultracentrifugation and microfiltration based on the protocol by Mutschelknaus et al. (17). Two centrifugation steps of 300 g and 10,000 g (4°C) were followed by 0.22 µm-filtration. The filtrate was centrifuged three times at 100,000g (4°C) for 2 h. Resuspended EV pellet was stored at -20°C. EV size distributions were analyzed by NanoSight LM10 (Malvern).

For vesicle transfer, 3 x 10⁵ BHY or FaDu cells were seeded in 6-well plates. After 24 h, the wells were washed with PBS and 2 ml fresh medium containing 10% EdFBS was added, supplemented by the EVs isolated from 6 ml conditioned medium from 0 or 6 Gy-irradiated cells. 24 h later, cells were detached with a cell scraper and analyzed via Western Blot.

To confirm EV-uptake into recipient cells, the isolated EVs were stained with green fluorescent dye PKH67 (MINI67, Sigma-Aldrich) as previously described (19). 40,000 BHY cells were seeded in 12-well silicone grids (Ibidi) on glass slides, after 24 h the wells were washed with PBS and 250 µl new EdFBS-supplemented medium was added, containing PKH67-stained EVs isolated from 750 µl conditioned medium of BHY cells. After 24 h, cells were fixed in 4% paraformaldehyde and nuclei were stained with Hoechst 33342.

For visualization of surface GRP78 after irradiation, cells were seeded on glass slides and after 24 h irradiated with 0 and 6 Gy. Further 24 h later, cells were fixed with 4% paraformaldehyde and labelled with an anti-GRP78 antibody dilution of 1:100 (PA1-014A, Invitrogen) followed by a secondary Alexa Fluor 488-coupled antibody (anti-rabbit: A-11008, Thermo Fisher). Cells were not permeabilized for appropriate labelling of surface GRP78.

Cell viability assay was performed with the PrestoBlue Cell Viability assay (Invitrogen) according to the manufacturers protocol.

For GRP78 inhibition, 5,000 cells were seeded in 96-well plates. After 24 h, medium was replaced with HA15- or DMSO-supplemented medium and cells were incubated for 24-72 h.

For GRP78 overexpression, 5,000 transfected cells (see Radiation Increased Surface Expression of GRP78 in HNSCC Cell Lines) were seeded in 96-well plates 48 h after transfection and incubated for 24-96 h.

Bioinformatic analyses including statistics were performed with R version 4.0.4 (27) and R Studio version 1.4.1106 (28). The package ‘ggplot2’ (29) was used for plots unless stated otherwise. For statistical analysis, the package ‘rstatix’ was used (30).

The data shown depict the mean +/- standard deviation of biological replicates (n). The significance level was chosen at 5%. For data with one factor variable, the paired, “two-sided” t-test was used and the Bonferroni correction was applied where appropriate. Remaining results were analyzed with 2-way repeated measures (RM) ANOVA and, when indicated, by pairwise “two-sided” t-tests with Bonferroni correction. Survival analysis was evaluated with log-rank test as mentioned in section 3.1.

To investigate the effect of GRP78 on treatment outcome of HNSCC patients, the TCGA database was used. GRP78 gene expression in tumor samples from 500 HNSCC patients was significantly higher than that in non-cancerous tissue samples obtained from 44 patients ( Figure 1A ). Analysis of the 43 available pairs of tumor and normal tissue samples confirmed this difference ( Figure 1B ). For survival analysis, data from 499 patients were divided into groups with low and high GRP78 gene expression, with the expression median chosen as the separation point. The results for the Kaplan-Meier analysis of survival revealed that patients with GRP78 RNA levels above the median had significantly shorter overall survival than patients with GRP78 RNA expression below the median ( Figure 1C ).

GRP78 protein expression in response to radiation was monitored in BHY and FaDu HNSCC cell lines. Neither of the irradiated cell lines revealed a change in GRP78 content when cells were detached with trypsin ( Figures 2A, C ). However, when cells were harvested by scraping, leaving surface proteins intact, both irradiated cell lines showed a significant increase in GRP78 expression ( Figures 2B, C ). Whilst radiation increased the trypsin sensitive pool of GRP78 on the surface of both BHY and FaDu cells, the amount of trypsin resistant intracellular GRP78 did not change The same results could be observed in the HNSCC cell lines CAL-33 and SCC131 ( Supplementary Figures S1A–C ).

To confirm the increase in cell surface expression, the scraped cells were analyzed by flow cytometry without permeabilization. This analysis corroborated the western blotting by showing a significant increase in antibody-accessible (surface) GRP78 expression in irradiated cells in comparison to the non-irradiated control cells ( Figures 2D, F and Supplementary Figure S1D ). Surface expression in cells gathered by trypsinization was not detectable ( Figure 2E ). In addition, increased surface expression of GRP78 after irradiation of BHY cells could also be shown with fluorescence microscopy ( Supplementary Figure S2 ).

GRP78 expression of both BHY and FaDu cells was significantly increased 48 and 72 h after transfection with the expression vector, versus cells transfected with empty control vector (pcDNA, Figures 3A–C ). Functional analyses were performed 72 h after transfection as GRP78 overexpression was higher at this timepoint.

To examine radioresistance, a clonogenic survival assay was performed. There was a significant increase in the radiation resistance of BHY and FaDu cells overexpressing GRP78 in comparison to vector control cells, embodied by a significant interaction parameter and an increase in the SF2 ( Figures 3D, E ).

Both GRP78-overexpressing cell lines also showed a significant increase in their ability to migrate into the cell-free gaps compared to vector control cells ( Figures 4A–D ). Cell viability was not affected by overexpression between 24 and 96 h compared to control transfection, with the 24 h timepoint being the start of cell migration assay ( Supplementary Figure S3A, B ).

The GRP78 inhibitor HA15 was used to block GRP78 function in BHY and FaDu cells. Cell viability analysis showed no significant effect of inhibitor treatment for both cell lines in comparison to control treatment with DMSO ( Supplementary Figures S3C, D ). Analysis of cell migration showed a dose-dependent decrease in cell motility in both cell lines in comparison to DMSO control treatment ( Figures 5A–D ). An analysis of the radiosensitivity was not possible as the drug treatment proved toxic due to low cell numbers and extended culture times needed for colony growth.

EVs derived from irradiated HNSCC donor cells have previously shown changes in their cargo and their ability to increase both the migratory potential and radioresistance of recipient cells (17, 19).

The identity of small EVs recovered from culture supernatants was confirmed by analysis of the particle size, showing a diameter of approximately 130 nm on average, the expression of the EV marker proteins Alix, TSG101 and CD9 and the absence of ER/cytosolic marker proteins Calnexin and GAPDH ( Figures 6A–C ). Analysis of total GRP78 content of EVs released from irradiated cells (6Gy-EVs) displayed greater amounts of GRP78 in comparison to EVs derived from the respective non-irradiated control cells (0Gy-EVs) ( Figures 6D, E and Supplementary Figures S1E, F ). Elevated levels of GRP78 were also found on the surface of 6Gy-EVs in comparison to 0Gy-EVs. Comparison of radiation-induced changes in GRP78 located on the EV surface showed a significantly greater change in expression of GRP78 relative to that of EV surface marker CD63. CD63 surface expression did not change significantly after donor cell irradiation ( Figures 6F, G ).

To analyze the effect of EVs derived from irradiated (6Gy-EVs) and non-irradiated HNSCC cells (0Gy-EVs) on the GRP78 expression of non-irradiated recipient cells, both cell lines were co-cultivated with the respective EVs. The addition of 6Gy-EVs onto non-irradiated recipient BHY or FaDu cells led to these cells showing increased amounts of GRP78 in comparison to 0Gy-EVs ( Figures 7A, B ). To verify that isolated EVs were incorporated by recipient cells, 0Gy-EVs were stained with fluorescent membrane dye PKH67, co-cultured with recipient cells and analyzed with fluorescence microscopy ( Figure 7C ).