Prostate cancer DU-145 and PC3 cell lines were purchased from the American Type Culture Collection (ATCC; Washington, DC, NW). DU-145 cells were cultured in Eagle’s minimum essential medium (EMEM) media (ATCC 30-2003) and PC3 cells in F-12K media (ATCC, 30-2004). For each cell line, the medium was supplemented with 10% heat-inactivated fetal calf serum (FCS; Merck KGaA, Darmstadt, Germany), and cells were cultured at 37°C in 5% CO₂. Cabozantinib (IPSEN, Paris, France) was dissolved in 1N HCl and sterile water (2.5 mg/ml).

Human-monocyte-derived DCs were generated from peripheral blood mononuclear cells (PBMCs) of healthy donors (Ethical Committee Protocol, Policlinico Umberto I—“Sapienza” University of Rome, Rif.5282/08.04.2019) as previously described (30). Monocytes (CD14⁺) were purified from PBMCs after Ficoll–Hypaque gradient (1,077 g/ml; Pharmacia LKB) using the Human CD14-Positive Selection Kit (StemCell Technologies, Vancouver, Canada) and cultured (5 × 10⁵ cells/ml) in Roswell Park Memorial Institute (RPMI)-1640 (Sigma-Aldrich) supplemented with 2 mmol/L L-glutamine, penicillin of 100 U/mL, streptomycin of 100 μg/ml (Sigma-Aldrich), with 5% heat-inactivated FCS (Merck KGaA). rhGM-CSF (50 ng/ml) (R&D Systems, Minneapolis, MN, USA) and 2,000 U/ml rhIL4 (R&D Systems) were added at day 0 and 2. Immature DCs (iDCs) were collected on day 5 and matured with cytokine cocktail [rhIL1β, IL6, TNFα (10 ng/ml) all purchased from R&D Systems and PGE₂ (1 μg/ml) purchased from Sigma-Aldrich)] for 16 h. Cabozantinib (2.5 μg/ml) was added at day 4 during DCs differentiation.

DC phenotype was analyzed by flow cytometry. Briefly, DCs were harvested, washed in phosphate-buffered saline (PBS) and resuspended at 10⁶ cells/ml. Each experimental sample was of 2 × 10⁵ cells/tube (BD Biosciences, NJ, USA), cells were incubated with specific fluorochrome-conjugated antibodies for 30 min at 4°C in the dark. After washing in PBS (two times), the cells were analyzed. The following monoclonal antibodies (mAb) were used: anti-HLAII-DR-APCH7 and anti-CD86-PeCy7, from BD Biosciences; anti-CD14-BB700, anti-CCR7-AlexaFlour647, anti-CD83-Pe, anti-CD40-BB515, anti-ICOS-L-Pe from BioLegend (San Diego, CA); and anti-VEGFR-1-PE from R&D Systems. MoAbs anti-IgG₁-BB515, -PE, -Pecy7, and -BB700; -AlexaFluor647; and -APC-H7 (BioLegend) were used as isotype controls. All mAbs were purchased from BD Biosciences and BioLegend. Flow cytometry analysis was performed using FACSCanto II flow cytometer running FACS Diva data acquisition and analysis software (BD Biosciences).

DU-145 and PC3 cells were seeded (2.5 × 10⁴ cell/ml) in 96-well plates (Corning Incorporated, New York, USA) and allowed to adhere overnight. Cells were then treated with serial dilution of Cabozantinib (2.5 and 5 μg/ml) for 24 and 48 h. At the end of each time point, the MTT assay (Roche Diagnostics, Basel, Switzerland) was performed according to the manufacturer’s instructions. Absorbance was measured at 550 nm.

Cells were seeded in six-well plates (Corning Incorporated; 2.5 × 10⁴ cell/ml) and left to adhere overnight. Cells were treated with serial dilution of Cabozantinib (2.5 and 5 μg/ml) for 24 and 48 h. As control, untreated cells were used. At the end of each time point, cells were harvested by trypsinization (1×, Sigma-Aldrich), resuspended at 10⁶ cells/ml in 1× Annexin V Binding Buffer (BD Biosciences). Cells were stained with 7-AAD and Annexin V-FITC (BD Biosciences) for 15 min to assess cell death. Flow cytometry was performed using FACSCanto II flow cytometer running FACS Diva data acquisition and analysis software (BD Biosciences).

Cells were seeded in a six-well plate (Corning Incorporated; 2.5 × 10⁴ cell/ml) and allowed to adapt overnight. Cells were then treated with serial dilution of Cabozantinib (2.5 and 5 μg/ml) for 24 and 48 h. The cells were fixed in 70% cold ethanol and incubated at 4°C overnight. The cells were incubated with RNaseA (Sigma-Aldrich) for 30 min and stained with propidium iodide (PI) (BD Pharmingen, San Diego, CA, USA). Flow cytometry was performed using FACSCanto II flow cytometer running FACS Diva data acquisition and analysis software (BD Biosciences).

iDCs, mDCs, DU-145, and PC3 cells were lysed using radioimmunoprecipitation assay (RIPA) buffer (1×, 100 μl/1 × 10⁶ cells, Cell Signaling, Beverly, MA, USA) supplemented with protease inhibitors (1×, Sigma-Aldrich) for 30 min on ice and then centrifuged at 13,000 g for 10 min.

The nuclear and cytoplasmic extracts were obtained from cells with and without serial dilution of Cabozantinib (2.5 and 5 μg/ml) at different time points (24 and 48 h). The cells were resuspended in Buffer A [20 mM HEPES, pH 7.9, 20 mM KCl, 3.0 mM MgCl₂, 0.3 mM Na₃VO₄, and freshly added 200 μM leupeptin, 10 mM E64, 300 μM phenylmethylsulfonyl fluoride (PMSF), 0.5 μg/ml pepstatin, 5 mM dithiothreitol (DTT), and 0.1% Nonidet P-40] and vortexed. After 30 min on ice, cells were centrifuged for 10 min at 10,000 g at 4°C. The supernatants were centrifuged another time for 30 min (10,000 g at 4°C), and the supernatants were taken as cytosolic extracts and frozen. To obtain nuclear extracts, pellets were resuspended in buffer B [40 mM HEPES, pH 7.9, 0.84 M NaCl, 0.4 mM ethylenediaminetetraacetic acid (EDTA), 50% glycerol, 0.3 mM Na₃VO₄, and freshly added 200 μM leupeptin, 10 μM E64, 300 μM PMSF, 0.5 μg/ml pepstatin, and 5 mM DTT] and vortexed. After 1 h on ice, nuclear extracts were cleared at 10,000 g for 1 h, at 4°C, and supernatants were transferred to fresh vials. Protein content was quantified by Bradford assay using bovine serum albumin (BSA) as a standard (Bio-Rad Lab, CA, USA), and samples were aliquoted and stored at −80°C.

Equal amounts of cell lysates or extracts, resuspended in sample buffer (Thermo Fisher Scientific, CA, USA) were resolved using 4%–12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gel (Thermo Fisher Scientific) and transferred to nitrocellulose membrane. After blocking with Tris-buffered saline with Tween 20 (T-BST) 5% milk (SERVA Electrophoresis Gmbh, Heidelberg, Germany), membranes were incubated with the following antibodies at a concentration 1:1,000: rabbit anti-β-catenin, anti-AXL, anti-Met, anti-β-tubulin, anti-HSP70, anti-Beclin1, and anti-mTORC all from Cell Signaling; rabbit anti-HMGB1, anti-Calreticulin, and anti-p62 were purchased from Abcam, Cambridge, UK; and anti-β-actin (Sigma-Aldrich, 1:10,000) and mouse anti-LC3 (MBL, Woburn, MA, USA). Membranes were incubated by peroxidase-conjugated goat antirabbit immunoglobulin G (IgG) (H+L; Jackson ImmunoResearch Laboratories; 1:20,000); peroxidase-conjugated goat antimouse IgG (H+L; Jackson ImmunoResearch Laboratories, West Grove, PA USA; 1:20,000). Protein bands were detected with horseradish peroxidase (HRP)-enhanced chemiluminescence (ECL) (Advansta, CA, USA), following the manufacturer’s instructions. The density of protein bands was analyzed by Image J software and was normalized in terms of the average intensity of bands of each protein per the average intensity of bands of β-tubulin or β-actin.

PC3 and DU-145 cells were seeded in 24-well plates and allowed to adapt overnight. The cells were treated with serial dilution of Cabozantinib (2.5 and 5 μg/ml) for 24 and/or 48 h. Untreated cells were used as experimental control. Calreticulin (CRT) membrane exposure was evaluated by flow cytometry, using as primary antibody anti-calreticulin mouse antibody (Abcam, 1:100), washing cells (two times), and then incubating the cells with PE-conjugated antimouse IgG (Southern Biotech Limited, USA). Flow cytometry was performed using FACSCanto II flow cytometer running FACS Diva data acquisition and analysis software (BD Biosciences). ATP release in the supernatant of cells was measured by means of an ENLITEN ATP Assay kit (Promega, Fitchburg, WI, USA), based on the ATP-dependent luciferin conversion, which yields detectable bioluminescence, according to the manufacturer’s protocol. HMGB1 concentrations in the supernatant of cells were measured by means of an enzyme-linked immunosorbent assay (ELISA) kit (TECAN, Zürich, Switzerland), according to the manufacturer’s protocol.

Extracellular vesicles (EVs) were purified from the cell culture supernatant of DU-145 cells. To isolate EVs, cells were cultured 3 × 10⁵ cells/ml in EMEM (ATCC) complemented with 2% fetal bovine serum (FBS) (Euroclone) for 24 or 48 h, with or without Cabozantinib (2.5 and 5 μg/ml). The supernatant underwent serial centrifugation steps at 4°C (550 g for 30 min, 1,500 g for 30 min). The supernatants were transferred in a fresh tube and then ultracentrifuged at 100,000 g for 1 h, at 4°C (Type 35 rotor, Beckman Coulter, USA). After the last ultracentrifugation step, the supernatants were discarded, and the final pellet containing EVs was gently resuspended in PBS without Mg⁺⁺ and Ca⁺⁺ (100 μl/pellet; Sigma-Aldrich), aliquoted and stored at −80°C. Protein concentration was measured by Bradford assay (Bio-Rad Laboratories). Analysis indicated that, in the presence of Cabozantinib, the optimal time point to obtain adequate EV preparation was at 48 h.

The uptake of DU-145 released antigens by DCs was first investigated. Briefly, DU145 cells were labeled with carboxyfluorescein succinimidyl ester (CFSE) (1 μM; Life Technologies Thermo Fisher, CA, USA) and incubated for 20 min at 37°C protecting from light. After two washes in PBS (37°C), the cells were treated with 2.5 μg/ml of Cabozantinib (48 h). Cell culture supernatants were collected and added to monocyte-derived DCs at the immature stage for 48 h. At the end of the incubation, the iDCs samples were acquired by a FACSCanto II flow cytometer (Becton Dickinson) and analyzed by FACSDiva software (Becton Dickinson) to verify the transfer of CFSE to the cytoplasm of the cells.

To study the endocytic capacity of DCs cultured in the presence or in the absence of 2.5 μg/ml Cabozantinib, FITC-dextran (1 mg/ml; Molecular Probes, Eugene, USA) was added to DCs for 1 h at 37°C and 4°C. After washing, cells were acquired by FACSCanto II flow cytometer (Becton Dickinson) and analyzed by FACSDiva software (Becton Dickinson).

Statistical analysis was performed using GraphPad Prism version 8 (Graphpad Software, Inc., San Diego, USA). Descriptive statistics [average and standard error media (SEM)] was used to describe the various data. Student’s paired t-test was used to compare two groups. Fold change represents the ratio between values obtained at treated and not treated cells (T/NT). Statistical significance was indicated when the p-value was <0.05.

The expression of the tyrosine kinase receptors (TKRs) targeted by Cabozantinib was investigated in the DU-145 and PC-3 prostate cancer cell lines. While the normal prostate glandular epithelium expresses low protein levels of AXL and cMET, the cell lines employed in the study displayed distinct profile of TKRs as detected by Western blot and/or flow cytometry and confirmed by public protein expression database (https://www.ebi.ac.uk/gxa/home) ( Supplementary Table S1 ). MET and AXL were expressed by both cell lines; in particular, AXL expression was more pronounced in DU-145 cells. Low expression levels of KIT and VEGFR were detected in DU-145 and PC-3 cells, respectively.

To test the impact of Cabozantinib on cell proliferation, cancer cells were cultured in the presence of increasing dose of the drug [2.5 μg/ml, as steady-state plasma concentration achievable in humans (31–33) and 5 μg/ml as exceeding dose], and results are shown in Figure 1A . Untreated (NT) cells were used as control.

Cabozantinib induced significant proliferative decrease in DU-145 cells at both drug concentrations and time points tested (at 24 h: p < 0.01 NT vs. 2.5 μg/ml and p < 0.05 NT vs. 5 μg/ml; at 48 h: p < 0.05 NT vs. 2.5 and 5 μg/ml). The response of PC-3 cells to Cabozantinib displayed a different kinetics: at 24 h, a significant proliferative decrease was observed only at the high dose of the drug (5 μg/ml). After 48 h, PC-3 proliferation significantly decreased at both drug concentrations (p < 0.01 NT vs. 2.5 and 5 μg/ml).

Induction of apoptosis did not appear to make account for the reduced cell growth observed upon Cabozantinib treatment. Indeed, no significant cell death induction was observed in PC-3 cells, independently by the treatment, and in DU-145 cells treated with 2.5 μg/ml of Cabozantinib. Apoptosis was only observed in DU-145 cells at the high drug dose, at both time points (p < 0.01) ( Figure 1B ).

These results prompted us to investigate whether Cabozantinib could restrain cell proliferation by cell cycle perturbation. As shown in Figure 1C , Cabozantinib treatment triggered cell cycle arrest in G1 phase, in both cell lines. The G1 phase arrest was pronounced at both Cabozantinib concentrations in both cell lines (p < 0.01) in the first 24 h, and this effect was maintained in DU-145 at 48 h ( Supplementary Figure S1A ). Concurrently, a decrease in S and G2 phases was observed at different extent in both the prostate cancer cell lines ( Supplementary Figure S1A ).

The unbalancing of the cell cycle may result in a strong cellular stress. HMGB1 is a nuclear protein: its cytoplasmic translocation is associated with cellular stress, and its release in the extracellular space indicates severe cell damage.

Therefore, we evaluated whether Cabozantinib could trigger immunogenic cell death (ICD) of DU-145 and PC-3 defined as HMGB1 release associated with simultaneous release of ATP and calreticulin (CRT) surface exposure. Following Cabozantinib treatment, HMGB1 increased in the cytoplasm fraction of DU-145 cells as detected by WB analysis ( Figure 2A ), and its extracellular release was detected in both cell lines by ELISA assay ( Figure 2B ).

This event was particularly enhanced in the DU-145 cell line at both Cabozantinib concentrations (24 h, p < 0.01; 48 h, p < 0.001). In the PC-3 cell line, significant HMGB1 release was observed at 24 h of treatment with the higher dose of 5 μg/ml (p < 0.01; Figure 2B ).

ATP release, a second feature of ICD, was also significantly induced by Cabozantinib in both cell lines at 48 h (DU-145: p < 0.05 NT vs. 2.5 μg/ml, p < 0.01 NT vs. 5 μg/ml; PC-3: p < 0.01 NT vs. 5 μg/ml) ( Figure 2C ). Then, the CRT exposure was evaluated: both cell lines did express baseline levels of CRT as often observed in cancer cells ( Figure 3 ). However, CRT exposure on cell membrane significantly increased in association with Cabozantinib concentration and treatment period (DU-145: p < 0.05 at 24 h for both 2.5 and 5μg/ml; p < 0.01 NT vs. 5 μg/ml at 48 h; PC-3: p < 0.01 NT vs. 2.5 μg/ml, p < 0.0001 vs. 5 μg/ml at 24 h; p < 0.05 NT vs. 2.5 and 5 μg/ml at 48 h).

The detection of the combined ICD markers following Cabozantinib exposure strongly suggests that the drug, although not inducing a cytotoxic effect on prostate cancer cells, impacted the quality of the apoptosis due to the release of ICD markers, with a marked effect on DU-145 cells.

Interestingly, DU-145 cells express high levels of AXL, which is targeted by this TKI. AXL has been shown to be crucial for modulating the balance between ICD and autophagy: AXL overexpression triggers macro-autophagic flux, while its inhibition arrests autophagy, inducing ICD (34). Indeed, Cabozantinib treatment induced upregulation of mTOR in DU-145 cells (negative at baseline) and decreased in the autophagic markers LC3-II and Beclin-1 ( Supplementary Figure S1B , p < 0.05).

These results suggest that Cabozantinib triggers immunogenic cell death and may offer a source of immunogenic tumor antigens.

The release of EVs mediates the crosstalk among cells in the tissue microenvironment and distant body districts (35). To verify whether tumor EVs contributed to the release of ICD molecular signals, EVs were isolated by ultracentrifugation from the culture supernatant of DU-145 cells grown without or with Cabozantinib (2.5 and 5 μg/ml) for 48 h, which resulted the optimal time point to obtain adequate EV preparation from cells grown in the presence of Cabozantinib.

EVs were analyzed by WB for the expression of ICD-associated molecules HMGB1 and CRT and the stress-associated protein HSP70, which was constitutively expressed in DU-145 cells, independently by TKI treatment ( Supplementary Figure S1C ).

HMGB1, CRT, and HSP70 were detected in the EVs shed by DU-145 cells upon Cabozantinib incubation. CRT appeared to be released associated with EVs also in the untreated cells and increased after treatment ( Figure 4A ); indeed, a baseline expression of CRT was detected in DU-145 cells ( Figure 3 ). These results suggest that the damage-associated molecules are released by prostate cancer cells in the microenvironment also by EVs.

To verify whether Cabozantinib-treated tumor cells could be a source of tumor antigens for antigen-presenting cells as such as dendritic cells (DCs), the DU-145 prostate cancer cells were labeled by CFSE and grown in the absence or presence of 2.5 μg/ml Cabozantinib (48 h). The cell culture supernatants were collected and added to monocyte-derived DCs at the immature stage (iDCs) for 48 h. At the end of the incubation, CSFE uptake by iDCs was analyzed by flow cytometry to verify the transfer of CFSE-labeled tumor antigens to DCs ( Figure 4B ). The percentage of CFSE⁺DCs increased after incubation with supernatants derived from CFSE-labeled tumor cells. This percentage further increased when tumor cells were treated with Cabozantinib. These results suggest that tumor cells were able to transfer antigens to DCs probably by the release of EVs and that this uptake increased after the addition of Cabozantinib to tumor cells.

We have shown in previous work that DCs biology is modulated by TKI as such as Pazopanib altering the Wnt/β-catenin pathway (10). Indeed, DCs do express VEGFR during their differentiation and may express MET after maturation (data not shown). The addition of Cabozantinib in DC culture during differentiation did not affect their viability (data not shown). We evaluated whether Cabozantinib could affect DCs analyzing the β-catenin signaling and phenotype changes, employing the drug concentration of 2.5 μg/ml.

The Wnt/β-catenin pathway stirs the activatory or tolerogenic capacity of DCs: the dampening of this pathway correlates with activation of DCs. Results demonstrated that Cabozantinib reduced the β-catenin pathway in iDCs. This reduction becomes statistically significant after maturation, suggesting the generation of activating mDCs (p < 0.01; Figure 5A ). This metabolic reprogramming corresponded to a phenotype modulation of DCs ( Figure 5B ). Among the DC marker analyzed, the expression of CD14, ICOSL, and CCR7 was significantly modulated by Cabozantinib. In particular, the CD14 molecule, expressed by monocytes and macrophages, was completely downregulated in mDCs (p < 0.05), while the costimulatory ICOS-L was strongly increased in the iDCs (p < 0.01). Interestingly, the CCR7 marker, a chemokine receptor that confers the ability to DCs to migrate towards the lymph nodes, was upregulated both in iDCs and mDCs (p < 0.05 and p < 0.01, respectively), suggesting that Cabozantinib sustained the activation of DCs. The expression of the other DC markers such as HLA-DR, CD83, CD40, and CD86, was not altered after the addition of Cabozantinib.

To understand whether Cabozantinib modulated the endocytosis function of DCs, the uptake of the fluorescein isothiocyanate (FITC)-dextran by DCs (treated or untreated) was evaluated ( Figure 5C ).

Data were reported as the ratio between the MFI from positive (dextran uptake obtained after 1 h at 37°C) and negative (dextran uptake after 1 h at 4°C) samples. Results showed a trend (p = 0.08) in the reduction in the endocytic capacity of DCs treated with Cabozantinib compared to the not treated cells, suggesting a more mature status of the cells cultured in the presence of the drug.