Stromal cell‐derived small extracellular vesicles enhance radioresistance of prostate cancer cells via interleukin‐8‐induced autophagy

Abstract

Radiation is a curative treatment for localized prostate cancer (PCa). Unfortunately, radiotherapeutic efficacy is often diminished when patients develop more aggressive or metastatic phenotypes. Recent studies have demonstrated that extracellular vesicles participate in cancer therapeutic resistance by delivering small bioactive molecules, such as small non‐coding RNAs. Here, we show that stromal cell‐derived small extracellular vesicles (sEVs) facilitate the radioresistance of PCa cells by transporting interleukin‐8 (IL‐8). Indeed, prostatic stromal cells secrete more IL‐8 than AR‐positive PCa cells, which can be accumulated in sEVs. Intriguingly, the uptake of stromal cells‐derived sEVs by radiosensitive PCa cells enhanced their radioresistance, which could be attenuated by silencing CXCL8 in stromal cells or inhibiting its receptor CXCR2 in PCa cells. sEV‐mediated radioresistance has been validated in zebrafish and mouse xenograft tumours. Mechanistically, the uptake of stromal sEVs triggers the AMPK‐activated autophagy pathway in PCa cells under the irradiation condition. Consequently, inactivating AMPK efficiently resensitized radiotherapy either by utilizing an AMPK inhibitor or silencing AMPKα in PCa cells. Furthermore, chloroquine (CQ), a lysosomal inhibitor, sufficiently resensitized radiotherapy via blockade of autophagolysosome fusion, leading to autophagosome accumulation in PC cells. Collectively, these results suggest that stromal cells enhance the radioresistance of PCa cells mainly through sEVs that deliver IL‐8.

1. INTRODUCTION

Prostate cancer (PCa) is a frequently diagnosed malignant tumour and the leading cause of cancer‐related deaths in men (Sung et al., 2021). Although radiotherapy can effectively preclude primary localized PCa (Aggarwal et al., 2021; Schaeffer et al., 2021), many patients with advanced high‐risk PCa eventually develop intrinsic or acquired resistance to radiation (Kratochwil et al., 2020; Mukha et al., 2021b), which remains an intractable obstacle that needs to be subdued urgently. The determinants of radiotherapy mainly rely on the generation of massive amounts of reactive oxygen species (ROS), which kill tumour cells by triggering multiple cell death pathways, such as necrosis, pyroptosis, apoptosis, and ferroptosis (Lei et al., 2020; Woo et al., 2020). Thus, the activation of antioxidants protects tumour cells from radiotoxicity (Josson et al., 2006). Conversely, inhibition of the cellular antioxidant defence system has been well‐documented in the sensitization of tumour cells to radiation (Zhang et al., 2018).

Autophagy, initially characterized as a primordial degradation pathway induced by various stresses, is essential for removing unnecessary or dysfunctional components through a lysosome‐dependent regulatory mechanism (Levine et al., 2019). Under metabolic stress conditions, AMP‐activated protein kinase (AMPK) adaptively provokes autophagosome formation by activating ULK1 or inhibiting mTORC1 preceding autolysosome‐mediated degradation (Dikic & Elazar, 2018). To date, autophagy is widely recognized as a conserved homeostatic process adapted to stress; therefore, autophagy deficiency leads to numerous pathological progressions, including cancer (Mizushima & Levine, 2020). Additionally, increasing evidence has shown that autophagy is a prevailing factor in cancer therapeutic resistance. Defective autophagy has been shown to reverse multidrug resistance in breast cancer (Ding et al., 2016). Furthermore, a recent study showed that the intervention of glutamine metabolism combined with the inhibition of autophagy resensitized PCa to radiation (Mukha et al., 2021a).

Activation of the cytokine/chemokine signalling pathways in tumour cells is thought to play a substantial role in the development of radioresistance. The TNF‐α/NF‐κB, IL6/STAT3, and TGF‐β/Smad2/3 signalling pathways have been well documented in tumour radioresistance (Yu et al., 2017). Additionally, other members of the interleukin family also participate in cancer radioresistance via activation of the NF‐κB pathway, including IL‐1, IL‐4, and IL‐10 (Wang et al., 2020). Indeed, the robust influence of cytokine signalling within the tumour microenvironment is essential for tumour cell vitality. In this regard, cancer‐associated stromal cells, including adipocytes, fibroblasts, pericytes, and macrophages, are a prerequisite for tumour metastasis and therapeutic resistance. Stromal cell‐derived cytokines/chemokines are necessary to support cancer progression and therapeutic resistance (Quail & Joyce, 2013; Sahai et al., 2020). Mounting evidence has demonstrated that cancer‐associated fibroblasts (CAFs) promote the radioresistance of cancer cells by transferring active molecules, including soluble cytokines (Zhang et al., 2019a, 2017). The CAF‐mediated radioresistance is mainly caused by activating the prosurvival signalling pathways, including NF‐κB and Notch (Huang et al., 2021; Nandi et al., 2022).

Interleukin‐8 (IL‐8, or CXCL8) is a proinflammatory chemokine that can bind to CXCR1 and CXCR2 on the cell surface and activate downstream signalling pathways in the cell (Alfaro et al., 2017; Ha et al., 2017). IL‐8 can be secreted by many cell types and is particularly stimulated by environmental stresses via multiple inflammatory signalling pathways, including ROS and hypoxia (Waugh & Wilson, 2008). As a critical mediator associated with inflammation and innate immune response, IL‐8 has been shown to correlate with angiogenesis, tumorigenesis, metastasis, and therapeutic resistance in various cancers (Fousek et al., 2021). We and others have previously shown that NF‐κB upregulates IL‐8 in advanced PCa, leading to enhanced radioresistance (Wang et al., 2020; Xu et al., 2009). In addition, increased IL‐8 secretion has also been characterized in endothelial cells, infiltrating neutrophils, immortalized epithelia, tumour‐associated macrophages, and fibroblasts, suggesting that IL‐8 may serve as a crucial oncogenic factor within the tumour microenvironment (Zhang et al., 2019b).

Small extracellular vesicles (sEVs) including exosomes, cell‐secreted nanoscale bilayer membrane structures, serve as vital cargos for intercellular communication within the tumour microenvironment. sEVs encapsulate intracellular substances, such as non‐coding RNA and proteins (Witwer & Thery, 2019; Yang et al., 2018). The molecules carried by sEVs can be transferred to the surrounding cells, thereby altering the phenotypes of recipient cells (Adamo et al., 2019; Wortzel et al., 2019). For instance, M2 macrophage‐derived exosomes transferred miR‐21 into gastric cancer cells and conferred cisplatin resistance in recipient cells through activating the PI3K‐AKT pathway by inhibiting PTEN (Zheng et al., 2017). In addition to miRNA, we recently reported that sEVs transfer adriamycin resistance by delivering Hsp70 into sensitive breast cancer cells, resulting in altered energy metabolism by switching mitochondrial respiration to glycolysis (Hu et al., 2021).

The present study shows that sEVs isolated from prostatic stromal cells with high constitutive IL‐8 levels efficiently enhance the radioresistance of AR‐positive PCa cells by delivering IL‐8. sEV‐transported IL‐8 induces AMPK‐mediated autophagy in recipient cells, and CQ efficiently resensitizes PCa cells to radiotherapy by inhibiting autophagy. The finding suggests that stromal cell‐derived sEVs enhance the radioresistance of PCa cells via the delivery of IL‐8 and activation of autophagy.

2. MATERIALS AND METHODS

2.1. Cell culture and treatment

Human AR‐positive PCa LNCaP and 22Rv1 cell lines, AR‐negative PCa PC‐3 and DU‐145 cell lines, human prostatic stromal myofibroblast WPMY‐1 cell line, and human prostatic viral immortalized epithelial RWPE‐1 cell line were purchased from American Type Culture Collection (ATCC, Manassas, VA, USA). PCa cell lines were cultured in RPMI 1640 medium (Gibco, Grand Island, NY, USA), and WPMY‐1 and RWPE‐1 cell lines were cultured in Dulbecco's modified Eagle's medium (DMED) medium, supplemented with 10% foetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco) in a humidified atmosphere of 5% CO2 at 37°C. PCa cells were cocultured with sEVs prior to IR treatment using an X‐ray machine (Rad Source RS2000 X‐ray, USA).

2.2. Isolation of cancer‐associated fibroblasts (CAFs)

Prostate tumour tissues were collected from patients newly diagnosed with PCa before chemotherapy or radiotherapy. The Ethics Committee of Nanjing Medical University approved the study protocol (2018‐565) with written informed consent obtained from the patients enrolled in this study. After washing with 1× PBS containing 100 U/mL penicillin (Gibco) and 100 μg/mL streptomycin (Gibco), the tissues were minced and then dissociated using collagenase IV (#2091, BioFroxx, Germany) containing 100 and 0.6 U/mL dispase II (#04942078001, Roche, Mannheim, USA) at 37°C for 30 min. The digested tissue mixture was filtered through a 75 μm filter (Millipore, Bedford, MA, USA), followed by centrifugation at 400 × g for 5 min to remove tissue debris. The suspended cells were sorted using anti‐fibroblast microbeads (#130‐050‐601, Miltenyi Biotec, Bergisch Gladbach, Germany). The collected CAFs were plated in 6‐well plates in RPMI‐1640 medium containing 10% FBS, 1% L‐glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin. The cultured CAFs were harvested and characterized using FACSCalibur Flow Cytometry (BD Sciences, San Jose, CA, USA) with primary antibodies against FAP (#66562, CST, Danvers, MA, USA) and α‐SMA (#69319, CST) with secondary antibodies conjugated with red or green fluorescent dye (#S0014, #S0018, Affinity). In addition, the CAFs morphological features were verified using microscopy.

2.3. Cell transfection

IL‐8 was silenced in WPMY‐1 cells by transfecting an shRNA plasmid targeting the CXCL8 gene (Merck, Billerica, MA, USA) using lipofectamine reagent (Invitrogen, Carlsbad, CA, USA). In addition, short hairpin RNA (shRNA) expression plasmids were used to silence AMPKα (Merck), CXCR2 (Merck), ATG5 (Tsingke, Nanjing, JS, China), and ATG7 (Tsingke) in PCa cells. Stable cell clones were obtained by puromycin selection. Sequences targeting CXCL8, GCTCTGTGTGAAGGTGCAGTT; AMPKα, GTTGCCTACCATCTCATAATA; CXCR2, CCGTCTACTCATCCAAT GTTA; ATG5, CCTTTCATTCAGAAGCTGTTT; and ATG7, TGAGTCATCAGT GGATCTAAA. In addition, we constructed a CXCL8‐3 × Flag in pCMV3‐SV40‐neo vector (#a2302024822, Corues, Nanjing, JS, China) and transfected it into endogenous CXCL8‐silenced WPMY‐1 cells. The stable cell clones were obtained by neomycin selection. The levels of IL‐8‐Flag in transfected cells and sEVs isolated from culture supernatants were confirmed by immunoblotting with antibodies against Flag (#14793S, CST) and IL‐8.

2.4. sEV isolation and purification

sEVs were isolated from the cell culture supernatants according to the updated MISEV2018 guidelines (Thery et al., 2018). Firstly, to remove serum EVs, FBS was centrifuged at 110, 000 × g at 4°C for 18 h using an ultracentrifuge (Beckman, Danvers, MA, USA). The medium was replaced with EV‐free medium when the cell density reached 30% confluence, and the cells were continuously cultured in EV‐free media for 2−3 passages. Approximately 8 mL culture medium was collected from the last subculture at 70%–80% cell confluence in a 100 mm culture plate. Only one‐time medium drawn was subjected to isolate sEVs using fractional centrifugation. Briefly, 50 mL of culture supernatant was collected from multiple plates, centrifuged at 500 × g for 15 min to remove the cell debris, and filtered through 0.22 μm filters (SLGPR33RB, Millipore, USA). The supernatants were spun down using gradient centrifugation at 2000 × g for 15 min, 5000 × g for 15 min, 12,000 × g for 30 min for removal of cell debris. EVs were precipitated by ultracentrifugation at 100,000 × g for 1 h at 4°C. The sEVs were further obtained from resuspended vesicles by continuous ultracentrifugation for 12 h to remove large‐scale EVs. Furthermore, to remove contaminated soluble components such as cytokines and signalling molecules, the sEVs (precip. sEVs) were finally purified by passing them through Amicon Ultra‐0.5 Centrifugal Filter Units, MWCO 100 kDa (#UFC510024, Millipore, Billerica, MA, USA) with a specific column for keeping sEVs. Finally, 1:5000 diluted precip. sEVs was analysed to determine the size and concentration of sEVs at 488 nm using a ZetaView Nanoparticle tracking analyser (NTA) (Particle Metrix, Meerbusch, Germany).

To collect total sEVs from cell culture supernatants, after removal of cell debris by centrifugation, the supernatants were filtered through 0.22 μm filters to remove particles, including large‐scale EVs (>220 nm). The filtered supernatants were passed through Amicon Ultra‐15 Centrifugal Filter Units, MWCO 30 kDa (#UFC903008, Millipore, USA) to concentrate sEVs (<220 nm), and then purified by passing through Amicon Ultra‐0.5 Centrifugal Filter Units, MWCO 100 kDa to trap 30–220 nm vesicles. Furthermore, the EV‐free supernatants (the soluble fraction) were concentrated by passing through Amicon Ultra‐0.5 Centrifugal Filter Units, MWCO 3 kDa (#UFC500396, Millipore). The concentrated total sEVs and soluble fraction were analysed by the NTA.

2.5. sEV characterization

The purified sEVs were fixed in 4% paraformaldehyde and 4% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), placed on a carbon‐coated copper grid, and immersed in 2% phosphotungstic acid solution for examination using TEM (JEOL Ltd., Tokyo, Japan) at an acceleration voltage of 80 kV. For sEV‐carried IL‐8 (sEV‐IL‐8) was quantified via immune electron microscopy using an IL‐8 primary antibody (#ab7747, Abcam, Cambridge, UK). In brief, the purified sEVs were embedded in 10% gelatin and soaked in a 2.3 M saccharose solution. Gelatin ultracryotomy (70 nm) was performed using a Leica FC7 (Leica, Vienna, Austria). The slides were incubated with 10 μg/mL IL‐8 antibody for 1 h at 4°C and then incubated with a goat anti‐rabbit IgG (10 nm gold) secondary antibody (#ab6712, Abcam) for 1 h. IL‐8 sorting into sEVs was examined using FEI Tecnai Spirit TEM (Hillsboro, OR, USA) at an accelerative voltage of 120 kV. To further characterize sEVs, the isolated sEVs were dissolved in lysis buffer and analysed via immunoblotting with specific antibodies against IL‐8 and EV markers, including CD63 (#sc‐365604, Santa Cruz Biotechnology, Dallas, TX, USA) and Tsg101 (#sc‐7964, Santa Cruz Biotechnology). An endoplasmic reticulum‐associated protein, calnexin, was also blotted using a relative antibody (#2433, CST) as a negative control for pure sEVs.

2.6. Human cytokine antibody array

sEV cytokine profiling was performed using a Bio‐Plex Pro Human Cytokine Screening Assay (#12007283, Bio‐Rad, Hercules, CA, USA) according to the manufacturer's instructions. This procedure was performed by Shanghai Hwayen Biotech., China. Briefly, 50 μg of sEV protein was subjected to the array and incubated with coupled magnetic beads. After washing twice, the array was incubated with biotin‐conjugated antibodies for 30 min and then with streptavidin‐phycoerythrin for 10 min. Signal was acquired using the by Bio‐Plex System (Bio‐Rad).

2.7. Protease protection assay

The purified sEVs were treated with 1× PBS containing 100 μg/mL of Proteinase K (# BS080, Biosharp, Shanghai, China) and 1% Triton X‐100 (Biofroxx) for 15 min, 37°C, and proteins were extracted using RIPA lysis (Beyotime, Shanghai, China). Proteinase K was inactivated by incubating the extracts at 95°C for 5 min, and proteins were separated using SDS‐PAGE.

2.8. Colocalization of sEV‐IL‐8 in the recipient cells

Purified sEVs were labelled with a red fluorescent dye PKH26 (#PKH26GL, Sigma, St. Louis, MO, USA) for 5 min. After washing twice with 1 × BSA to remove excess dye, the stained sEVs were recovered by ultracentrifugation. 5.0 × 10⁴ cells were seeded in confocal dishes and then cocultured with 1.9 × 10¹⁰ particles/mL PKH26‐labelled sEVs for 12 h, fixed in ice‐cold methanol for 10 min, and permeabilized with 0.1% Triton X‐100 for 10 min. After blocking with 5% BSA for 30 min, the cells were stained with an IL‐8 antibody for 16 h and then incubated with a secondary antibody containing a green fluorescent dye (#S0018, Affinity, Nanjing, JS, China) for 1.5 h. CD63 was also stained to colocalize with the IL‐8 image. The nuclei were counterstained with DAPI. Fluorescence was visualized and captured using a confocal microscope (LSM800, Carl Zeiss Jena, Oberkochen, Germany). The colocalization of florescent images was quantified by Image J software (National Institutes of Health, Maryland, USA). To quantify immunofluorescence (IF), the colocalization of CD63‐IL‐8 in the selected area uptake index was computed by determining the total colocalized puncta or macropinosome area in relation to the cytoplasmic area (at least three fields). The average intensity of CD63 or IL‐8 was calculated by the “grey value” feature by ZEN2 software (Carl Zeiss Jena GmbH, Germany), as indicated by arbitrary unit (a.u.) and the percentage of colocalization (Co‐l, %).

2.9. Localization of IL‐8‐CXCR2 on the cell membrane

To verify the interaction between stromal IL‐8 and CXCR2 of the recipient cells, CXCL8‐Flag was ectopically expressed in IL‐8‐silenced WPMY‐1 cells, and stromal sEVs carried IL‐8‐Flag were cocultured with 22Rv1 cells for 24 h. Cytoplasmic fraction of the recipient cells was isolated using a Cytoplasmic Protein Extraction Kit (#0028, Beyotime), and the uptake of sEV‐IL‐8‐Flag by the recipient cells was confirmed by immunoblotting using a Flag antibody. Furthermore, immunoprecipitation (IP) was performed to determine IL‐8‐Flag binding to CXCR2 using a protein A/G magnetic Co‐IP kit (#AM001‐01, ACE, Nanjing, China). After uptake of sEVs, 22Rv1 cells were treated with 2 Gy IR. Proteins extracted from the cells were incubated with anti‐Flag or anti‐IgG magnetic beads (#P2115, P2171, Beyotime) at 4°C for 16 h. After washing the beads thrice, the Flag‐IP or IgG‐IP was eluted from the magnetic beads. The eluted proteins were detected by immunoblotting with antibodies against Flag, IL‐8, and CXCR2.

For localization of sEV‐IL‐8 on the cell membrane of the recipient cells, PCa cells were seeded in six well plates at a density of 4 × 10⁵ cells/well overnight. The cells were co‐cultured with 3.8 × 10¹⁰ particles/mL sEVs with the Flag‐tag overnight. The media were removed and washed with 1 × PBS twice. After adding the fresh medium, the cells were irradiated with 2 Gy IR. One, two, or four hours after irradiation, the cells were fixed in 4% paraformaldehyde for 20 min, and then incubated with APC‐conjugated anti‐Flag antibody (#637307, Biolegend, San Diego, CA) for 30 min. After washing with 1 × PBS thrice, the fluorescent signal was analysed using a FACSCalibur Flow Cytometry (BD Sciences, San Jose, CA, USA).

In addition, for localization of sEV‐transported IL‐8‐Flag to CXCR2 of the membrane of the recipient cells, after sEV‐IR combined treatment, the cells were incubated with a Flag antibody (#66008‐4‐Ig, Proteintech) and CXCR2 antibody (#20634‐1‐AP, Proteintech) and followed by incubation with red or green fluorescence‐conjugated secondary antibody (#SA00013‐4, #SA00013‐1, Proteintech). The dual fluorescence was visualized and captured using a confocal microscope (Zeiss, Germany), and the percentage of colocalization of IL‐8 and CXCR2 was quantified by Image J software.

2.10. Radiation‐induced cytotoxicity

PCa cells were seeded into 96‐well plates at a density of 1.0 × 10⁴ cells/well overnight. After cocultured with 3.8 × 10¹⁰ particles/mL sEVs for 24 h, the cells were irradiated using 2 Gy ionizing radiation (IR). To inhibit the effect of stromal sEVs on IR‐mediated cytotoxicity, stromal cell culture was pretreated EV inhibitor (GW4869, #HY‐19363, MCE, New Jersey, USA). In addition, PCa cells were pretreated with multiple agents for 1 h before IR to manipulate the effect of IL‐8. To block the effect of stromal IL‐8 on tumour cells, PCa cells were treated with anti‐IL‐8 antibody (#CSD00668, HuMax‐IL‐8, Chemstan, Wuhan, HB, China) or an inhibitor of CXCR2 (AZD‐5069, #HY‐19855, MCE). To verify the effect of IL‐8 on radioresistance, PCa cells were pretreated with a recombinant Human IL‐8 (#208‐IL‐010, Novus Biologicals, Minneapolis, USA). To inhibit IL‐8‐activated AMPK pathway, PCa cells were pretreated with an AMPK inhibitor (BAY‐3827, #HY‐112083, MCE). To prevent IL‐8‐induced autophagy, PCa cells were pretreated with chloroquine (CQ) (#HY‐17589A, MCE). After 24 h, the cell viability was quantified using a CCK‐8 kit (#K1018, APExBIO, Houston, TX, USA). Furthermore, cell survival rate was quantified using a colony survival assay. Cells were seeded in six well plates at 200 cells/well and maintained for 12 h. After treatment, the cells were incubated for 14 days to allow colony formation. The colonies were stained with 1% crystal violet and counted. The cell survival rate was calculated as the number of colonies normalized to the cell plating efficiency.

2.11. Quantification of autophagic cells

PCa cells were cultured in serum‐free medium for at least three passages. To examine autophagy, cells were seeded in confocal dishes at 5 × 10⁴ cells/well overnight in serum‐free media. The medium was replaced with serum‐free medium containing 1.9 × 10¹⁰ particles/mL sEVs for 24 h and then treated with 2 Gy IR. Three hours after irradiation, cells were fixed in ice‐cold methanol for 10 min and permeabilized with 0.1% Triton X‐100 for 10 min. After blocking with 5% BSA for 30 min, the cells were stained with primary antibodies against LC3 and Lamp2 (#14600‐1‐AP, #66301‐1‐Ig, Proteintech, Wuhan, HB, China) for 16 h and then incubated with a secondary antibody with a green or red fluorescent dye (Affinity) for 2 h. The nuclei were counterstained with DAPI. The fluorescence was visualized and captured using a confocal microscope (Zeiss, Germany) and quantified by Image J. In addition, the colocalization of autophagosomes with lysosomes was doubly imaged by merging the LC3 image with the Lamp2 image. The colocalization of fluorescent images was quantified by Image J as described in 2.8.

2.12. Blockade of autophagy by CQ

PCa cells were seeded in six well plates at a density of 1 × 10⁵ cells/well overnight. The cells were infected with a lentiviral LC3 with green and red dual fluorescence (#GM‐1314L204H, Genomeditech, Shanghai, China) in Opti‐MEM medium containing 3 μg/mL polybrene for 48 h (MOI = 10). The stable LC3 expression cell clones were selected using puromycin. To block the autophagy process, PCa cells were pretreated with 10 μM CQ before sEV‐IR combined treatment. To monitor the treatment‐induced autophagic process, the fusion of autophagosomes and autophagolysosomes was doubly imaged by confocal microscopy. The effect of CQ on the blockage of autophagolysosomes was estimated using the ratio of red to green fluorescence.

2.13. Lysosomal acid phosphatase assay

Lysosomal acid phosphatase activity was measured using an Acid Phosphatase Kit (#P0326, Beyotime). After treatment, cells were harvested, lysed in 10% Triton X‐100 and then centrifuged at 12,000 rpm for 10 min. Lysosomal acid phosphatase activity was measured in the supernatant according to the manufacturer's instructions.

2.14. RNA isolation and RT‐qPCR

After coculture with stromal sEVs for 24 h, 22Rv1 cells were treated with 2 Gy IR and continuously cultured for 24 h. Total RNA was extracted from treated cells using TRIzol reagent and subjected to 2% agarose gel electrophoresis. RNA (2 μg) was converted to cDNA using the PrimeScript RT reagent Kit (#RR037Q, Takara, Kusatsu, Shiga, Japan). cDNA (20 ng) was used to perform RT‐qPCR using a Taq Pro Universal SYBR qPCR Master Mix (#Q712‐02, Vazyme, Nanjing, China) and quantified using a LightCycler System (Roche, San Francisco, CA, USA). The relative expression level of CXCL8 mRNA was quantified by normalizing to GAPDH mRNA. Sequences of the specific PCR primers for amplifying CXCL8 cDNA: forward primer, 5′‐AACTGAGAGTGATTGAGAGTGG‐3′; reverse primer, 5′‐ATGAATTCTCAGCCC

TCTTCAA‐3′; and for GAPDH cDNA: forward primer, 5′‐TCTGACTTCAACAGC

GACACC‐3′, reverse primer, 5′‐CTGTTGCTGTAGCCAAATTCGTT‐3′.

2.15. Immunoblotting analysis

Total cellular proteins were extracted using a RIPA lysis reagent containing 1 mM phenylmethylsulphonyl fluoride (PMSF; Santa Cruz Biotechnology) and quantified using a BCA kit (Beyotime). Cell extracts (50–100 μg) were separated on 10% SDS‐polyacrylamide gels and then transferred to PVDF membranes. The membranes were incubated overnight at 4°C with the primary antibodies against LC3, Lamp 2, β‐actin (Proteintech), AMPKα (#07‐350, Millipore), phospho‐AMPKα (#50081, CST), mTOR (#2972, CST), and phospho‐mTOR (#2971, CST). The membranes were washed thrice with TBST buffer and incubated at 25°C for 2 h with an HRP‐conjugated secondary antibody (#sc‐2357, Santa Cruz Biotechnology; #7076, CST). Immunoblots were visualized using an enhanced chemiluminescence detection system (Bio‐Rad). The intensities of the blots were quantified using Quantity One software, and protein expression was normalized to loading controls such as β‐actin.

2.16. ELISA

The levels of IL‐8 in the cell culture media and sEVs were quantified using a human IL‐8 cytokine kit (#EHC008.96, Xinbosheng, Shenzhen, China) according to the manufacturer's protocol.

2.17. Cellular ATP measurement

The levels of ATP in the cell culture media and purified sEVs were quantified using an ATP Assay Kit (#S0027, Beyotime) according to the manufacturer's protocol.

2.18. In vivo validation using a zebrafish model

An experimental zebrafish tumour model was established to validate the results of the in vitro studies. 22Rv1 cells were cocultured with 3.8 × 10¹⁰ particles/mL sEVs derived from WPMY‐1 cells in FBS‐free medium for 3 days. After washing with 1× PBS, the cells were stained with CM‐Dil (UE, China) in FBS‐free medium for 5 min and then washed with 1× PBS thrice to remove excess dye. The cells were pretreated with CQ and then treated with IR. Two hundred live cells were microinjected into zebrafish embryos (aged 48 hpf, n = 10). The injected cells were attached to the yolk and spread evenly throughout the organism. The injected embryos were incubated at 28°C for 1 h and then turned to 34°C to continue the cultivation. —Seven to ten days after injection, the tumours formed in adult fish were imaged every other day using a fluorescent microscope at 583 nm, and the fluorescence intensity was quantified using ZEN2 software.

2.19. In vivo validation using a nude mouse model

Animal experiments were performed in accordance with the Institutional Animal Care and Use approved by the Research Committee of Nanjing Medical University (No. IACUC‐1901031). A mouse tumour xenograft experimental model was used to confirm the results obtained from the zebrafish tumour xenografts. 22Rv1 cells were cocultured with 3.8 × 10¹⁰ particles/mL sEVs derived from WPMY‐1 cells with different levels of IL‐8 for 3 days and the tumour cells were continuously cocultured with stromal sEVs for three passages. Five‐week‐old athymic BALB/c nude male mice (n = 5) (Beijing Vital River Lab Animal Tech. Co., Ltd., China) were subcutaneously injected with 5 × 10⁶ (in 100 μL) sEV‐pretreated 22Rv1 cells. In addition, sEVs (3.8 × 10¹⁰ particles in 100 μL) were subcutaneously injected twice weekly after tumour implantation before irradiation treatment. Mouse weight and tumour volume were measured every other day. After the tumour volume reached 300 mm³, mice were treated with localized IR (3 × 3 Gy administered given every 4 days). The mice were sacrificed as tumour growth reached the maximal volume (2000 mm³) in the no IR treatment group, and tumour tissues were excised for immunoblotting, H&E, and IHC.

2.20. Haematoxylin‐eosin staining (H&E) and immunohistochemistry (IHC)

Tumour tissues were fixed in paraffin‐embedded sections and then dewaxed with xylene. For H&E, the tissue sections were stained with haematoxylin‐eosin. For IHC, the tissue sections were soaked in 5% BSA buffer for 1 h and then incubated with 400× diluted primary antibodies in 5% BSA buffer at 4°C overnight, including against 400× diluted Ki‐67 (#27309‐1‐AP, Proteintech) 500× diluted caspase3 antibody (#9662, CST). After washing with 1 × PBS, the tissue sections were incubated with an HRP‐conjugated secondary antibody at room temperature for 1 h. After washing with 1×PBS, DAB Substrate Kit (Cell signalling Tech) was used to stain the tissue sections. Additionally, the tissue sections were incubated with 100× diluted LC3 or Lamp2 antibody and followed by incubating with green or red fluorescence‐conjugated secondary antibody and washed with 1× PBS. The immunostaining images were observed by Digital Pathology System (Pannoramic MIDI, 3DHISTECH, Budapest, Hungary). The expression levels of relative proteins in the tissues were quantified using Image J software. The intensity of IHC staining was analysed as previously described.

2.21. Bioinformatics analysis

The Cancer Genome Atlas (TCGA) database (https://www.cancer.gov/about‐nci/organization/ccg/research/structural‐genomics/tcga was used to assess the correlation between IL‐8 and PCa progression. In addition, the PCa tumour single‐cell transcriptome dataset GSE157703 in the Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/ query/acc.cgi?acc = GSE157703) was analysed to estimate the constitutive levels of IL‐8 in AR‐positive and stromal cells. Furthermore, a dataset of PCa‐associated cell lines GSE19426 in GEO (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE19426) was analysed to forecast IL‐8 expression profiles in PCa cell lines versus stromal and viral immortalized cell lines. Finally, a dataset of the radiosensitivity of the PCa cell lines GSE80657 in GEO (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi) was analysed to predict the signalling pathways involved in radioresistance.

2.22. Statistical analysis

Data are presented as the mean ± standard deviation (SD) from at least three replicates. Significant differences between experimental groups were analysed using unpaired Student's t‐test. One‐way analysis of variance (ANOVA) followed by Dunnett's or Bonferroni's multiple comparison tests was performed using Prism (GraphPad, San Diego, USA). Statistical significance was set at p < 0.05.

3. RESULTS

3.1. Stromal cells enhance the radioresistance of PCa cells with a low level of IL‐8

Previously, we demonstrated that RelB, an NF‐κB member, contributes to the development of castration‐resistant phenotypes characterized by increased IL‐8 and decreased PSA (Xu et al., 2009). To assess the supporting role of IL‐8 in PCa therapeutic resistance, we analysed a prostate tumour single‐cell transcriptome dataset from the GEO database. First, CXCL8 expression profiles in AR‐positive PCa and stromal cells were defined. Particularly, a high abundance of IL‐8 was identified in PCa‐associated stromal cells compared to AR‐positive PCa cells (Figure S1a‐c). In addition, we also searched a dataset of PCa‐associated cell lines in the GEO database, and the results confirmed that constitutive IL‐8 in PCa‐associated stromal cells was higher than that in primary AR‐positive PCa cells. However, IL‐8 expression was significantly increased in AR‐negative metastatic PCa cells (Figure S1d, e).

Furthermore, we examined IL‐8 mRNA and protein levels in PCa cell lines. Compared to stromal cells, LNCaP and 22Rv1, the typical AR‐positive PCa cell lines sensitive to radiotherapy expressed low levels of IL‐8 (Figure 1a–c). LNCaP cells were cocultured with the supernatant of WPMY‐1 cells to examine whether stromal cell components enhance PCa cell radioresistance. As expected, the cell viability and survival rate were apparently increased due to stromal cell stimulation, compared to no coculture or coculture with PCa cell supernatant (Figure 1d–f). Likewise, 22Rv1 cell radioresistance was also enhanced after coculture with stromal cell supernatant (Figure 1g–i). To verify whether stromal IL‐8 contributes to PCa radioresistance, an IL‐8 therapeutic antibody was used to eliminate the effect of stromal IL‐8 (Figure 1j). To test whether EV cargos delivered IL‐8 from stromal cells to tumour cells, stromal cell culture was pretreated with an EV inhibitor (GW4869) to deprive the effect of stromal EV. As expected, the inhibitor sufficiently abrogated the stromal effect on the radioresistance of tumour cells (Figure 1k). These results suggest that stromal cells contribute to the radioresistance of PCa cells by transporting stimulatory factors, including IL‐8.

FIGURE 1.

Stromal cell supernatant enhances the radioresistance of PCa cells. (a–c) The expression levels of the CXCL8 gene in human PCa, prostatic stromal myofibroblast and viral immortalized prostatic epithelial cell lines were quantified using RT‐qPCR (n = 3) (a) and immunoblotting (b). The cell‐secreted IL‐8 in cell cultural supernatants was measured using an IL‐8 ELISA kit (n = 4) (c). (d–f) LNCaP cells were cocultured with the supernatants (labelled as “s”) derived from stromal WPMY‐1 cells or their supernatants or left no supernatant as a control before IR treatment. The cytotoxicity was examined by CCK‐8 assay (n = 3) (d) or colony survival assay (n = 3) (e,f), respectively. (g‐i) In addition, 22Rv1 cells were also cocultured with the supernatants and then treated with IR. The cell survival rate was also quantified (n = 3). (j) Furthermore, after cocultured with stromal supernatant, 22Rv1 cells were pretreated with an IL‐8 therapeutic antibody (Humax‐IL‐8) before IR treatment. The cell survival rate was examined using a CCK‐8 assay (n = 5). (k) Stromal cells were pretreated with an EV inhibitor GW4869 (10 μM) or DMSO vehicle for 48 h. Stromal supernatants were collected and cocultured with 22Rv1 cells before IR treatment. The cell survival rate was examined using a CCK‐8 assay (n = 6). *(p < 0.05), **(p < 0.01) and ***(p < 0.001) show significances between the different groups as indicated; ns shows no significance.

3.2. Stromal cell‐derived sEVs transported IL‐8 into PCa cells

To test whether sEVs serve as cargos for intercellular delivery of active molecules, sEVs were isolated from the culture supernatants of WPMY‐1, LNCaP and 22Rv1 cells according to the standard procedure described in the MISEV2018 guidelines (Thery et al., 2018). Morphological characteristics of sEVs were examined under TEM, which revealed lipid bilayer membrane structures with diameters of <200 nm in size (Figure 2a). Since cytokines are widely recognized as crucial factors for cancer radioresistance acquisition (Wang et al., 2020; Xu et al., 2009), the purified sEVs were subjected to a human cytokine antibody assay to quantify cytokine profiling in sEVs. Although most cytokines were lower in sEVs derived from stromal cells than sEVs from PCa cells, several important cytokines were apparently higher in stromal sEVs than those in tumorous sEVs (Figure S1f). Notably, IL‐8 was highly enriched in stromal sEVs compared to that in tumorous sEVs (Figure S1g).

FIGURE 2.

Stromal sEVs transport IL‐8 into the recipient PCa cells. (a) sEVs were isolated from cultural supernatants of PCa and stromal cells. The structure of the isolated sEVs was examined using TEM as indicated by red arrows. (b) Furthermore, sEVs‐carried IL‐8 (sEV‐IL‐8) was quantified by immune TEM using an IL‐8 antibody, as indicated by dark particles. (c) The characters of the isolated sEVs were examined using immunoblotting with specific antibodies against EV membrane‐specific proteins, including CD63 and Tsg101. Additionally, sEV‐IL‐8 was also quantified as a sEV‐carried protein, and calnexin serves as a cytosolic protein control. Furthermore, the filtrates taken from the purification of sEVs by specific columns were also blotted to control potentially contaminated proteins adhered to sEVs during sEVs preparation. (d) The isolated stromal sEVs was subjected to protease K, and CD63, Tsg101, and IL‐8 were quantified by immunoblotting. (e) The concentration in sEV‐ IL‐8 was quantified using an IL‐8 ELISA kit (n = 6). (f, g) LNCaP and 22Rv1 cells were cocultured with the isolated sEVs or left with no coculture as a control. The cell uptake of sEVs was examined by confocal microscopy using a CD63 antibody with red fluorescence and an IL‐8 antibody with green fluorescence. Colocalization of sEVs and IL‐8 in the recipient cells was indicated by merged fluorescent signals with a DAPI image for localizing nuclei. The merged signals were amplified from small areas indicated by white boxes. The fluorescent intensity (IL‐8, CD63, a.u) was quantified by ZEN2 software and the percentage of IL‐8‐CD63 colocalization (Co‐l, %) were calculated using Image J software. (h) After sEV‐cell coculture, LNCaP cellular extracts were immunoblotted with an IL‐8 antibody to determine the increased cell‐internalized IL‐8. (i) In addition, the concentration of IL‐8 in cultural supernatants was measured using an IL‐8 ELISA kit (n = 6). (j,k) Additionally, the cell‐internalized IL‐8 in 22Rv1 cells uptake of sEVs was also quantified by immunoblotting and ELISA. ***(p < 0.001) shows the significance between different groups as indicated; ns shows no significance.

Subsequently, the purified sEVs were incubated with an IL‐8 antibody and sEV‐IL‐8 was detected using immune TEM. IL‐8 was encapsulated in sEVs, and practically more abundant IL‐8 was found in sEVs derived from stromal cells than in PCa cells (Figure 2b). To control the purity of sEVs, the isolated sEVs were further verified by immunoblotting using specific antibodies against the EV markers CD63 and Tsg101. In parallel, the cytoplasmic protein calnexin was used to control the cytosolic protein contamination of sEVs. To control the purification of sEVs, filtrates that flowed out from specific sEVs ultrafiltration columns were blotted to examine any contaminated cytosolic proteins that may attach to sEVs during the ultracentrifugation procedure (Figure 2c). To verify intact sEVs, the purified sEVs were subjected to a protease protection test. Protease K was able to digest CD63, an EV membrane protein, but failed to degrade Tsg101 and IL‐8 in sEVs, unless pretreating with Triton X, a membrane‐permeable agent (Figure 2d). In addition, IL‐8 levels in the purified sEVs were quantified using its ELISA kit. Consistently, the IL‐8 concentration in stromal cell‐derived sEVs was much higher than that in PCa cell‐derived sEVs (Figure 2e).

To assess the efficiency of sEVs isolation by ultracentrifugation, the isolated sEVs and total sEVs were analysed by the NTA. The sizes of sEVs were between 50 and 250 nm (>90%) (Figure S2a). Approximate 20% sEVs were precipitated, and EV was undetectable in the soluble fraction by NTA (Figure S2b). Additionally, the distribution of IL‐8 was analysed by quantifying IL‐8 in a single stromal cell versus relative culture supernatant containing EVs and then quantifying IL‐8 located in total sEVs, precip. sEVs, and free IL‐8 in the soluble fraction. The secreted IL‐8 in the medium was more than in the cell. Approximate 50% IL‐8 was free in the soluble section, and 40% IL‐8 was encapsulated in total sEVs, including 15% IL‐8 in precip. sEVs (Figure S2c). Additionally, IL‐8 was encapsulated in precip. sEVs and no IL‐8 existed in filtrates from the sEVs purification columns (Figure S2d).

To determine the effect of each fraction on radioresistance, 22Rv1 cells were cocultured with stromal cell culture supernatant, total sEVs, precip. sEVs, and EV‐free soluble fraction derived from the same volume of supernatants before IR treatment. Although the effect of precip. sEVs was lower than the supernatant and total sEVs, the effect of precip. sEVs was higher than the soluble fraction; particularly, 5× precip. sEV equal to the number of total sEVs provided the maximal protective response (Figure S2e, f). To verify the observation, 22Rv1 cells were cocultured with the same amount of IL‐8 from precip. sEVs or the soluble fraction for 0–24 h before IR treatment. The protective effect of the soluble fraction was increased at 6 h and then turned to decreases. Conversely, the effect of precip. sEVs was highly increased at 24 h (Figure S2g). Additionally, the distribution of IL‐8 in the recipient cells and culture media was analysed. Although IL‐8 in media was decreased in the two groups, IL‐8 was remarkably increased in the recipient cells uptake of sEVs compared to the soluble fraction, indicating that IL‐8 is unstable in the soluble fraction (Figure S2h, i). These results suggested that sEVs facilitate IL‐8 stability and delivery to the recipient cells.

To examine whether sEVs from stromal cells can transport IL‐8 into PCa cells, sEVs were labelled with the red fluorescent dye PKH26 before coculture with PCa cells and IL‐8 was imaged using its antibody with green fluorescence. sEVs were engulfed by the cells and IL‐8 was internalized in the recipient cells; in particular, the amount of IL‐8 was highly increased in the uptake of stromal cell‐derived sEVs compared to the uptake of their own sEVs (Figure S3a, b). Additionally, the EV marker CD63 was imaged using its antibody with red fluorescence, which was merged with the IL‐8 green fluorescence image. The results confirmed that uptake of stromal sEVs increased IL‐8 in PCa cells (Figure 2f, g). Moreover, IL‐8 concentrations in PCa cells before and after uptake of sEVs were quantified by immunoblotting and ELISA. Coculturing with stromal sEVs increased IL‐8 concentration in the culture media of PCa cells. Intriguingly, 24 h after the sEV‐cell coculture, the level of IL‐8 increased in PCa cells as it decreased in the media (Figure 2h–k). These results suggest that stromal sEVs can deliver IL‐8 from stromal cells to tumour cells within the tumour microenvironment.

3.3. Stromal sEVs deliver IL‐8 and enhance the radioresistance of PCa cells

To examine whether stromal sEVs enhance the radioresistance of the recipient PCa cells, 24 h after the sEV‐cell coculture, the PCa cells were treated with IR and cell viability was determined using a CCK‐8 assay. The results indicated that stromal sEVs protected PCa cells from radiotoxicity in the dose‐dependent manners (Figure 3a–c). Additionally, to test whether IL‐8 is a major contributor to stromal sEVs‐mediated radioresistance, PCa cells were treated with IL‐8 pure protein before IR treatment. Consistently, IL‐8 dose‐dependently protected PCa cells from radiotoxicity (Figure 3d). In addition, the sEV‐cocultured cells were treated with dose‐escalated IR, and radiotoxicity was quantified using clonogenic assay. The cell survival rate was remarkably increased by the uptake of stromal sEVs (Figure 3e–h).

FIGURE 3.

Stromal sEVs enhance the radioresistance of PCa cells. (a,b) PCa cells were cocultured with stromal sEVs (3.8×10¹⁰ particles/mL) or their sEVs or left no coculture as a control. Cytotoxicity was determined using a CCK‐8 assay (n = 3). (c, d) PCa cells were cocultured with dose‐escalated stromal sEVs or IL‐8 protein followed by IR treatment. The cytotoxicity was quantified using a CCK‐8 assay (n = 5). (e,f) After uptake of sEVs, LNCaP cells were treated with dose‐escalated IR as indicated. The cell survival rate was determined using a colony survival assay by normalized to cell plating efficiency (n = 3). (g,h) Additionally, after the sEV‐IR combined treatment, the survival of 22Rv1 cells was also determined (n = 3). **(p < 0.01) and ***(p < 0.001) show the significances between the different groups as indicated; ns shows no significance.

In addition, PCa cells were cocultured with sEVs isolated from viral immortalized prostatic epithelial RWPE‐1 cell with a high level of IL‐8, and then treated with IR. Consistently, the cell survival rate was increased by uptake of sEVs (Figure S4a–d). To verify that the protective effect of stromal sEVs on PCa cell survival relies on transporting IL‐8, but not on the contaminated soluble components obtained from the preparation of sEVs, filtrates flowed out from sEVs purification columns were incubated with PCa cells before IR treatment. The results indicated no radioprotective effect observed in the filtrate‐treated cells (Figure S4e–h). Additionally, instead of stromal sEVs, IL‐8 was used to treat PCa cells followed by IR treatment. Consistently, the results of colony survival assay confirmed the protective effect of IL‐8 against radiotoxicity (Figure S4i, j). Furthermore, to rule out the possibility of endogenous IL‐8 influencing the radioresistance, the relative mRNA level of CXCL8 in 22Rv1 cells was measured after treatment. The results showed no significant change in endogenous IL‐8 by sEV‐IR combined treatment (Figure S4k).

In addition, we examined whether the uptake of IL‐8 can bind to CXCR2 on the recipient cell membrane under the irradiation condition. Firstly, we ectopically expressed Flag‐tagged IL‐8 in endogenous IL‐8‐silenced WPMY‐1 cells (Figure S5a) and isolated sEVs from the cell culture supernatant (Figure S5b). After coculture with 22Rv1 cells. the uptake of stromal sEV‐IL‐8‐Flag by the recipient cells was confirmed by immunoblotting using a Flag antibody (Figure S5c). After sEVs‐IR combined treatment, IL‐8 receptor CXCR2 on the membrane of the recipient cells was immunoprecipitated using an anti‐Flag antibody (Figure S5d). In addition, IL‐8‐Flag localized on the cell membrane was quantified by flow cytometry. Intriguingly, the amount of IL‐8‐Flag increased at 1–2 h but decreased at 4 h after the combined treatment (Figure S5e–g). The results were further confirmed by confocal microscopy with double fluorescence to quantify the colocalization of IL‐8‐Flag and CXCR2 on the cell membrane. Consistently, the combined treatment increased the percentage of IL‐8‐Flag colocalized with CXCR2 on the cell membrane at the early time (Figure S5h–j).

Moreover, primary prostate tumour tissues from patients with PCa were used to isolate CAFs following a standard procedure. After purification by CAF‐specific microbeads, CAF primary cells were characterized to express CAF‐specific proteins (FAP and α‐SMA), whereas very few PCa‐specific proteins (AR and PSA) were detected (Figure 4a–c). Subsequently, sEVs were isolated from the CAFs culture media. The level of IL‐8 in CAF‐derived sEVs was comparable to previous sEVs derived from WPMY‐1 cell culture media (Figure 4d, e). Correspondingly, CAF‐derived sEVs were cocultured with PCa cells prior to IR treatment. Similarly, CAF‐sEVs also enhanced the radioresistance of PCa cells (Figure 4f–j).

FIGURE 4.

sEVs derived from CAFs enhance the radioresistance of PCa cells. (a) CAFs were isolated from prostate primary tumour tissues, and the specific phenotype was examined under microscopy. (b,c) The isolated CAFs were further verified by flow cytometry using antibodies against fibroblastic markers (α‐SMA and FAP) and PCa markers (PSA and AR). (d) The levels of IL‐8 in CAF‐derived sEVs and stromal cell‐derived sEVs were measured using its ELISA kit (n = 3). (e) sEVs were isolated from CAF cultural supernatant and characterized by immunoblotting. (f) CAF‐derived sEVs were cocultured with 22Rv1 cells before IR treatment. Cytotoxicity was quantified using a CCK‐8 assay (n = 3). (g‐j) Furthermore, the effect of CAF‐sEVs on PCa cell survival was confirmed using colony survival assay (n = 3). **(p < 0.01) and ***(p < 0.001) show the significances between different groups as indicated; ns shows no significance.

3.4. Stromal sEV‐IL‐8 provokes AMPK‐mediated autophagy in the recipient cells

To assess the mechanisms by which stromal IL‐8 enhances PCa cell radioresistance, we further analysed a PCa cell line GEO dataset using a pathway‐enriched approach. An atlas of multiple signalling pathways was identified in PCa radioresistant cell lines compared with PCa radiosensitive cell lines. Notably, the pathway involved in AMPK‐activated senescence/autophagy was correlated with IL‐8 abundance (Figure S6a, b). To verify this observation, after the sEV‐IR combined treatment, the levels of phospho‐AMPKα, the autophagic marker LC3 I/II, and the lysosome marker Lamp2 in PCa cells were quantified by immunoblotting. Compared to the untreated control, the uptake of stromal sEVs appeared to increase the levels of these autophagic markers, particularly 3 h after irradiation (Figure S6c, d).

Subsequently, to assess whether AMPK‐activated autophagy is involved in stromal sEV‐enhanced PCa cell radioresistance, the amount of ATP in sEVs uptake PCa cells was measured. Although slight changes were observed in IR‐untreated cells, the uptake of stromal sEVs significantly reduced ATP production at 3 h after irradiation but increased ATP at 6 h after irradiation (Figure 5a–d), indicating that the combined effect of sEV‐IR may trigger AMPK activation at the early time point. In parallel, the IL‐8 and IR combined treatment similarly modulated ATP production (Figure S7a–c). Correspondingly, silencing of CXCR2 prior to irradiation led to the reverse direction of ATP production (Figure S7d–f).

FIGURE 5.

sEV‐IR combined treatment induces autophagy in PCa cells. (a‐d) 3h or 6h after sEV‐IR combined treatment, the level of ATP in the treated PCa cells was quantified (n = 6). **(p < 0.01) and ***(p < 0.001) indicate the significances between the two groups; ns indicates no significance. (e, f) After sEV‐IR combined treatment, IL‐8 and autophagic markers in the cells were quantified by immunoblotting using specific antibodies. β‐actin serves as a loading control. (g, h) After the sEV‐IR combined treatment, autophagy in the cells was examined by confocal microscopy using LC3 and Lamp2 antibodies with green or red fluorescence, respectively. The autophagy formation in the cells was examined by merging fluorescent signals with a DAPI image for localizing nuclei; the merged signals were amplified from small areas indicated by white boxes. The percentage of LC3‐Lamp2 colocalization (Co‐l, %) was calculated using Image J software. (i,k) The structure of autophagy was further verified using TEM. The fusion of autophagosomes or autophagolysosomes in the cells was indicated by yellow arrows or red arrows, respectively.

Metabolic stress‐mediated adaptive AMPK activation is essential for inducing autophagosome (Dikic & Elazar, 2018). In this regard, we further examined the levels of autophagic proteins in PCa cells at 3 h after the sEV‐IR combined treatment. Notably, the level of phospho‐AMPKα was remarkably increased by the combination treatment. Accordingly, LC3‐II and Lamp2 were simultaneously elevated in sEV‐IR‐treated cells. Conversely, no such evidence was observed in PCa cells that treated with sEVs or IR alone. Since slight mTOR inhibition was detected in sEV‐IR treated cells, it is considered that AMPK plays a determinant role in stromal sEVs‐induced autophagy (Figure 5e, f). Furthermore, autophagic cells were doubly imaged by confocal microscopy using relative antibodies with red and green fluorescence. The merged LC3 and Lamp2 signals showed that only the combined treatment was sufficient to induce autophagy (Figure 5g, h). The structure of autophagy was further examined by TEM. Compared to the untreated control, the sEV‐IR combined treatment prominently promoted autophagosomes and autophagolysosomes (Figure 5i, j).

3.5. Inactivation of IL‐8 resensitizes PCa cells to IR

To determine whether stromal sEV‐IL‐8 is crucial for enhancing the radioresistance of PCa cells, IL‐8 was silenced in stromal cells using an shRNA targeting CXCL8 and confirmed in the cells and cell‐secreted sEVs (Figure S8a–c). PCa cells were cocultured with IL‐8‐deprived stromal sEVs prior to IR treatment. As expected, silencing of IL‐8 efficiently abolished the sEV‐mediated radioprotective effect (Figure 6a, b). In parallel, the levels of phospho‐AMPKα, LC3 and Lamp2 also decreased, indicating that IL‐8 is critical for the progression of autophagy under irradia‐tion conditions (Figure 6c). Consistently, the percentage of LC3‐Lamp2 colocalization was increased by the sEV‐IR combined treatment but the image was significantly alleviated by silencing IL‐8 in stromal cells (Figure 6d, e).

FIGURE 6.

Inhibition of IL‐8 and AMPK resensitizes PCa cells to IR by inhibiting autophagy. (a, b) IL‐8 was silenced in stromal cells using a shRNA targeting the CXCL8 gene. 22Rv1 cells were cocultured with sEVs isolated from the IL‐8‐deprived stromal cells and followed by IR treatment. The cell survival rate was quantified using colony survival assay (n = 3). (c) After the sEV‐IR combined treatment, the levels of IL‐8 and autophagic markers in 22Rv1 cells were quantified by immunoblotting with specific antibodies. β‐actin serves as a loading control. (d,e) Compared to the IR‐untreated cells (d), the formed autophagy in IR‐treated cells (e) was detected by confocal microscopy using LC3 and Lamp2 antibodies with green or red fluorescence, respectively. (f, g) AMPKα was further silenced in 22Rv1 cells using a shRNA approach and followed by the sEV‐IR combined treatment. The cell survival rate was quantified using colony survival assay (n = 3). (h) After sEV‐IR combined treatment, the levels of phospho‐AMPKα and autophagic markers in 22Rv1 cells were quantified by immunoblotting with specific antibodies. Two groups were separated by loading the protein molecular weight makers as indicated by ‘M’. β‐actin serves as a loading control. (i) The formed autophagy was detected by confocal microscopy using LC3 and Lamp2 antibodies with green or red fluorescence, respectively. The merged signals were amplified from small areas indicated by white boxes. The percentage of LC3‐Lamp2 colocalization (Co‐l, %) was calculated using Image J software. *(p < 0.05) and **(p < 0.01) show the significances between the two groups as indicated; ns shows no significance.

Furthermore, 22RV1 cells were pretreated with a CXCR2 inhibitor before the sEV‐IR combined treatment. The inhibition of CXCR2 in the cells uptake of sEVs efficiently resensitized radiotherapy by eliminating the effect of stromal sEV‐IL‐8 on the radioresistance (Figure S8d, e). Consistently, silencing of CXCR2 led to re‐sensitizing the cells to IR by reducing the effect of stromal sEVs (Figure S8f–h).

3.6. AMPK activation is involved in stromal sEV‐provoked autophagy

To verify that AMPK is critical for triggering stromal sEV‐IL‐8 induced autophagy, AMPKα was further silenced in PCa cells. The silence of AMPKα appeared to show no change in cell survival of unirradiated cells, irrespective of sEVs uptake. However, the cell survival rate was remarkably reduced by silencing AMPKα in the sEV‐IR treated cells (Figure 6f, g). In fact, AMPK silencing efficiently abrogated the activation of LC3 and Lamp2, irrespective of sEVs uptake, IR treatment, or their combination (Figure 6h, i). In addition, an inhibitor of AMPK was applied to verify that AMPK serves as the primary regulator of sEV‐IL‐8 induced autophagy under irradiation conditions. Stromal sEVs protected PCa cells against radiotoxicity, but this protective effect was alleviated by pretreatment of PCa cells with an AMPK inhibitor (Figure S9a, b). Consistently, the percentage of colocalization LC3‐Lamp2 image was reduced by pretreatment with the AMPK inhibitor (Figure S9c).

Furthermore, downstream regulators of autophagy were examined to confirm the activation of the autophagic pathway is involved in stromal sEV‐mediated radioresistance. ATG5 and ATG7 were silenced in PCa cells before sEV‐IR combined treatment (Figure S9d). Under unirradiated conditions, the defects in ATG5/ATG7 showed no cytotoxic effect, regardless of the presence of stromal sEVs. Intriguingly, ATG5/ATG7 dysfunction remarkably reduced sEV‐mediated radioprotection effect on tumour cell survival (Figure S9e, f). These results were consistent with the inactivation of IL‐8 and AMPK in sEV‐IR‐treated cells. Accordingly, the silence of ATG5/ATG7 abolished the sEV‐IR combined treatment‐induced LC3I/II and Lamp2 signals (Figure S9g–i). Collectively, these results suggest that stromal sEV‐IL‐8 protected PCa cells from radiotoxicity mainly via triggering AMPK‐mediated autophagy.

3.7. CQ blocks autophagy and resensitizes PCa cells to IR

CQ is a currently available clinical drug that ameliorates tumour chemotherapeutic resistance by inhibiting autophagy which functionally blocks the fusion of the autophagosome with lysosome by deacidifying lysosome (Levy et al., 2017). Thus, we applied CQ to counteract stromal sEV‐mediated radioresistance by inhibiting autophagy. First, the CCK‐8 assay was used to examine the cytotoxicity of CQ in 22Rv1 cells. The results indicated that the IC50 value of CQ was 16μM, and 10μM CQ efficiently resensitized the cells to IR (Figure S10a, b). Additionally, a colony survival assay was performed to further examine the inhibitory effect of CQ on stromal sEV‐mediated radioresistance (Figure 7a, b; Figure S10c, d). Mechanistically, CQ sufficiently blocked lysosome activation by decreasing its acid phosphorylation, resulting in increased LC3, indicating that CQ led to the accumulation of autophagosomes in the cells uptake sEVs (Figure 7c, d; Figure S10e, f). Furthermore, PCa cells were injected with a lentiviral LC3 probe labelled with dual fluorescence (LC3‐GFP‐RFP) to quantify autophagosomes and autophagolysosomes. The sEV‐IR combined treatment enhanced autophagosomes and then tended to autophagolysosomes, but CQ efficiently impaired the fusion of autophagolysosomes by decreasing the acidic compartment, leading to accumulated autophagosomes (Figure 7e, f). These results provide proof‐of‐concept evidence that CQ resensitizes PCa cells to IR by inhibiting lysosome‐linked autophagy.

FIGURE 7.

CQ resensitizes PCa cells to IR by inhibiting autophagolysosomes. (a, b) The sEV cocultured 22Rv1 cells were pretreated with 1μM CQ and followed by IR treatment. The cell survival rate was determined using colony survival assay (n = 3). (c) After pretreating cells with 10μM CQ, the levels of phospho‐AMPKα and autophagic markers in the cells were quantified by immunoblotting using the specific antibodies. β‐actin serves as a loading control. (d) The activity of lysosomal acid phosphatase in the cells was measured. (e, f) A lentiviral LC3‐GFP‐RFP was stably transduced into 22Rv1 cells. The sEV cocultured 22Rv1 cells were pretreated with 10 μM CQ and followed by IR treatment. The fusions of autophagosomes and autophagolysosomes were doubly imaged by confocal microscopy using a LC3 antibody. The merged images in the white boxes were amplified to show accumulated autophagosomes in the CQ‐treated cells. The fluorescent mean intensity (GFP‐LC3, RFP‐LC3, a.u.) and the ratio of red to green fluorescence was calculated using ZEN2 software. **(p < 0.01) and ***(p < 0.001) indicate the significances between two groups; ns indicates no significance.

3.8. Stromal sEV‐mediated radioresistance of PCa cells was monitored In vivo

To monitor the pathological aspect of stromal sEVs in sustaining PCa radioresistance, a zebrafish tumour xenograft model was established to imitate in vitro observations. sEVs isolated from WPMY‐1 cells were cocultured with PKH26‐labelled 22Rv1 cells with or without CQ in serum‐free medium. The cells were treated with IR and then microinjected into zebrafish embryos. Tumours were formed as the fish grew and screened using fluorescent imaging. As expected, the tumours rapidly grew in the IR‐untreated groups. Notably, although IR was sufficient to suppress tumour growth in the IR‐treated groups, sEVs could efficiently rescue tumour survival, and CQ dramatically resensitized the tumours to IR (Figure S11a, b).

In addition, a mouse tumour xenograft model was further used to validate whether stromal sEVs enhance PCa radioresistance via the delivery of IL‐8. 22Rv1 cells were pretreated with sEVs derived from stromal cells with different levels of IL‐8 and the sEV‐cell mixture was subcutaneously injected into 5‐week‐old nude male mice. To keep the stromal sEVs function, the purified sEVs were subcutaneously injected into mice twice weekly. The formed tumours were treated with IR when the tumour volume reached 300 mm³. IR sufficiently inhibited the tumour growth compared to that in the untreated control. Notably, sEVs remarkably protected tumour survival against radiotoxicity, but the protective effect was alleviated in cells pretreated with sEVs isolated from IL‐8‐deprived stromal cells (Figure 8a–c). Subsequently, the excised tumour tissues were examined to quantify the relative levels of autophagic markers. Compared to the no sEVs uptake control, the levels of IL‐8, phospho‐AMPKα, LC3 and Lamp2 strikingly increased in the sEV‐IR combined treatment group, but the sEVs effect was dramatically alleviated by silencing IL‐8 in the stromal cells (Figure 8d). In addition, tissue slides were imaged to analyse the structure of autophagy. Consistently, autophagy was impressively stimulated by the combined treatment, but the effect of stromal sEVs on autophagy formation was diminished by silencing of IL‐8 (Figure 8e). The tumour proliferation index was further analysed using H&E and IHC staining. The results of H&E staining and Ki67 imaging indicated that stromal sEVs protect tumour cells from radiotoxicity mainly through the delivery of IL‐8. In parallel, caspase 3 image further revealed that sEVs could rescue cells from IR‐mediated apoptotic cell death by provoking the autophagic survival signalling pathway; however, the protective effect was further alleviated by silencing IL‐8 in sEVs (Figure 8f–h). Taken together, the finding of the present study demonstrates that prostatic stromal cells enhance the radioresistance of PCa cells via sEVs that transport IL‐8, as illustrated in Figure 8(i).

FIGURE 8.

Stromal sEV‐IL‐8 enhances the radioresistance of mouse xenograft tumours. (a) 22Rv1 cells were cocultured with sEVs derived from stromal WPMY‐1 cells with different levels of IL‐8. The mixture of sEV‐cell was subcutaneously injected into nude male mice (n = 5). Additionally, sEVs were continuously injected into mice twice weekly. The formed tumours were treated with 3 × 3 Gy IR started at the tumour volume reaching 300 mm³. After tumours reached the maximal volume (2000 mm³) in the IR‐untreated control groups, the mice were sacrificed and the excised tumour tissues were photographed. (b) Tumour volume was measured every other day, and the tumour growth rate was determined. (c) Accordingly, tumour weight was plotted in each group. (d) The levels of IL‐8, phospho‐AMPKα and autophagic markers in excised tumour tissues were quantified by immunoblotting using specific antibodies. β‐actin serves as a loading control. (e) Colocalization of autophagy in the tumour tissues was imaged by confocal microscopy using LC3 and Lamp2 antibodies with green and red fluorescence, respectively. (f‐h) The tumour tissues were further imaged using H&E staining, IHC against Ki67 and caspase 3. The images of Ki67 and caspase 3 and the relative H‐score values were quantified by Image J. *(p < 0.05), **(p < 0.01) and ***(p < 0.001) indicate the significances between different groups; ns indicates no significance. (i) Depiction of the suggested mechanistic integration involved in stromal sEV‐enhanced radioresistance of PCa cells.

4. DISCUSSION

Although the 5‐year survival rate of PCa has been steadily improving in Western countries, such as the USA, the morbidity and mortality of PCa are rapidly increasing in developing countries, including China (Miller et al., 2019). Radiotherapy is widely recognized as one of the most popular treatment options for localized PCa. However, many patients eventually develop advanced PCa that is resistant to conventional radiotherapy. Although these patients were treated with improved radiotherapy, PCa can still relapse after definitive radiotherapy due to both intrinsic and acquired radioresistance developed during radiotherapy (Chang et al., 2014). Thus, there is an urgent need to decipher the tumour environmental machinery underlying PCa radioresistance, thereby discovering novel interventional targets to improve PCa radiotherapeutic efficacy.

The tumour microenvironment is often similar to the soil where tumour cells can grow, which consists of tumour cells and other cell types, blood vessels, signalling molecules, and extracellular matrixes surrounding the tumour cells (Neviani et al., 2015). Apart from tumour progression supporting therapeutic resistance, the tumour environmental components such as chemoattractants secreted by stromal cells promote and sustain anticancer resistance (Baghban et al., 2020). Recent studies suggest that cancer‐associated stromal cells, such as fibroblasts, mesenchymal stem/stroma, osteoblasts, and chondrocytes, are essential for sustaining tumour cell growth and survival (Valkenburg et al., 2018). Of the stimulators generated by stromal cells, cytokines are crucial for their malignant potential and therapeutic resistance (Fiori et al., 2019). For instance, inhibition of TGF‐β in fibroblasts leads to enhanced anti‐PD‐L1 immunity for treating urothelial cancer by facilitating the penetration of CD8⁺ T‐effector cells into tumour cells (Mariathasan et al., 2018). Pancreatic stellate cell‐secreted IL‐6 fosters chemoresistance of pancreatic cancer cells by activating the Stat3 signalling pathway (Dosch et al., 2021). Notably, EVs serve as the important cargos to shuttle microenvironmental substances into tumour cells (Li & Nabet, 2019).

EVs can deliver active molecules from donor cells to recipient cells within the tumour microenvironment. Numerous studies have demonstrated that EV‐mediated transport of specific miRNAs affects therapeutic resistance by regulating the target genes in recipient cells (Shen et al., 2019). In addition, EVs can deliver protective proteins and offer chemotherapeutic resistance to drug‐sensitive tumour cells (Hong et al., 2020; Morrissey & Yan, 2020). In particular, a variety of cytokines were detected in EVs, including exosomes derived from cancer and cancer‐associated stromal cells, such as TGF‐β, TNF‐α, IL‐6, IL‐8, and IL‐10 (Tan et al., 2018). The present study further demonstrated that sEVs derived from stromal cells can enhance the radioresistance of AR‐POSITIVE PCa cells that express low levels of IL‐8 and are sensitive to radiotherapy. After endocytosing stromal sEVs, the recipient cells appeared to be resistant to IR via the activation of autophagy‐mediated survival pathways. Conversely, CQ‐mediated inhibition of autophagy re‐sensitized the cells to IR.

IL‐8, an 8.3 kDa small chemokine, its biological effects are mediated mainly by binding to G protein‐coupled cell surface receptors CXCR1/2 (Holmes et al., 1991). The high level of IL‐8 is correlated with tumour malignancy and therapeutic resistance in various cancers (Kuai et al., 2012; Ning et al., 2011; Park et al., 2014). IL‐8 contributes to advanced PCa with declined PSA via activation of NF‐κB and suppression of AR (Maynard et al., 2020; Xu et al., 2009). Consequently, targeting IL‐8‐mediated inflammatory signalling led to sensitizing PCa to anticancer agents, oxaliplatin and TRAIL (Waugh & Wilson, 2008). Notably, Lopez‐Bujanda et al. recently reported that castration increases IL‐8 production and promotes PCa progression via IL‐8‐promoted myeloid infiltration. Thus, targeting IL‐8 facilitated immune response via immune checkpoint blockade (Lopez‐Bujanda et al., 2021). Furthermore, this study showed that IL‐8 enhances the radioresistance of AR‐positive PCa cells via sEV‐mediated crosstalk between stromal cells and tumour cells.

It has been demonstrated that as a key inflammatory factor in the tumour microenvironment, IL‐8 promotes immune evasion by reducing PD‐L1 blockade (Olivera et al., 2022; Yuen et al., 2020). Exosomal IL‐8 from cancer cells has been shown to promote chemoresistance through crosstalk with stromal cells (Chen et al., 2019; Corrado et al., 2014). In addition, IL‐8 secreted from cancer‐associated stromal cells promotes metastasis and chemoresistance in multiple types of cancer (Le Naour et al., 2020; Markovina et al., 2010; Pausch et al., 2020). In addition to IL‐8, other cytokines conjugated with EVs activate relative receptors on the surface of the recipient cells, including TGF‐β and CCL2 (Lima et al., 2021; Ringuette Goulet et al., 2018). We have recently reported that adriamycin‐resistant breast cancer cell‐derived sEVs conferred resistance to the sensitive cells by delivering TGF‐β1 (Tan et al., 2021). The present study demonstrates that sEVs derived from prostatic stromal myofibroblasts enhance PCa radioresistance through the delivery of IL‐8 into tumour cells. Mechanistically, stromal IL‐8 protects tumour cells against radiotherapy by provoking the AMPK‐mediated autophagy pathway.

Nevertheless, a recent study has shown the distribution of cytokines conjugated on the surface of EVs and encapsulated in EVs. Since cytokines out of EVs can directly activate relative receptors on the cell membrane, this study suggested that EVs may not be necessary for signal transduction (Fitzgerald et al., 2018). The prediction was dependent upon the classical paradigm for cytokine‐mediated intercellular signalling. However, this study showed that sEV‐IL‐8 endocytosed by tumour cells could further migrate and activate CXCR2 on the cell surface under irradiation. In our model, engulfing EV was essential for sEV‐IL‐8 triggering AMPK signalling in the recipient cells. After sEV‐cell coculture, radiation increased the colocalization of IL‐8/CXCR2 on the recipient cell membrane. Dysfunction of CXCR2 in the recipient cells alleviated the biological effect of stromal sEVs.

Indeed, previous studies have demonstrated evidence that cell‐secreted cytokines can activate the receptors on the cell membrane (Hartman et al., 2013). Particularly, tumour cell‐produced IL‐8 promotes EMT and metastasis via activation of CXCR2 on the cell membrane (Chen et al., 2015; Long et al., 2016). Notably, a recent review article emphasized the uptake of exosomes can be re‐secreted from the recipient cells, which can be a path for consuming exogenous exosomes (Kalluri & LeBleu, 2020). In this regard, it has been well characterized that exosomal PD‐L1 function is dependent on the exosome isolation method (Shu et al., 2020). Thus, controversial observations may be generated from different procedures to isolate and purify EVs. The present study suggests that the current model exists in sEV‐mediated cell‐cell communication. Nevertheless, this study did not rule out the possibility of sEV‐IL‐8 directly activating CXCR2 on the cell membrane before or in the process of uptake by the recipient cells.

Autophagy is a lysosome‐mediated intracellular self‐catabolic degradation process for maintaining cellular homeostasis to determine the cell fate, either life or death (Doherty & Baehrecke, 2018; Mizushima & Komatsu, 2011). Autophagy regulates the prosurvival pathway mainly through the clearance of protein aggregates and damaged organelles to provide bioenergetic substances for survival (Kroemer & Levine, 2008). However, numerous studies have demonstrated that metabolic stress‐mediated autophagic cell death is provoked by the crosstalk between autophagy and apoptosis, ferroptosis, or necrosis (Dai et al., 2020; Kessel & Reiners, 2020; Nikoletopoulou et al., 2013). Although autophagy suppresses tumorigenesis in the early stage of the tumour, in most contexts, it fosters tumour growth by dealing with microenvironmental stress and maintaining cellular homeostasis (Mowers et al., 2017; White, 2015). Subsequent studies have shown that autophagy plays a pivotal role in cancer therapeutic avenues through metabolic remodelling, and thereby, the inhibition of autophagy enhances therapeutic efficacy (Gremke et al., 2020; Wang et al., 2018; Zhang et al., 2015).

It is now well‐admitted that AMPK promotes autophagy in response to starvation, hypoxia, and oxidative stress (Jiang et al., 2021; Karabiyik et al., 2021). Conversely, mTORC1 inhibits autophagy by maintaining nutritional status (Kamada et al., 2010; Kim et al., 2011). Furthermore, cytokine‐activated inflammatory signalling cascades promote autophagy to sustain tumour survival, invasion, and metastasis (Monkkonen & Debnath, 2018). The present study ascertained that triggering autophagic survival signalling is the definitive mechanism by which stromal cell‐derived sEVs enhance radioresistance of PCa cells by transporting IL‐8 into tumour cells. Notably, the activation of AMPK only occurred in the combined treatment of sEVs and IR, but the treatment slightly affects mTOR, suggesting that stromal IL‐8 promotes autophagy in tumour cells under irradiation conditions, mainly through stimulation of the AMPK‐ULK1 signalling axis.

In summary, sEVs are essential mediators of cell‐to‐cell communication. Stromal cells‐derived sEVs contribute to therapeutic resistance via delivery of active molecules within the tumour microenvironment. Since IL‐8 contributes to PCa progression and radioresistance, the implication of stromal sEV‐transported IL‐8 in triggering autophagic survival signalling is a major concern for acquired radioresistance in PCa cells. Thus, insights into sEV‐mediated IL‐8 intercellular transport are anticipated to provide a promising approach for discovering novel malignant biomarkers and therapeutic targets by capturing sEVs.

AUTHOR CONTRIBUTIONS

Xiumei Wang: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Resources; Validation; Visualization; Writing—original draft; Writing—review & editing. Fan Xu: Data curation; Formal analysis; Methodology; Resources; Software; Validation. Hengyuan Kou: Data curation; Formal analysis; Visualization. Yawen Zheng: Resources; Software; Writing—original draft. Jing Yang: Resources; Software; Writing—original draft. Zhi Xu: Data curation; Formal analysis; Methodology; Validation. Yao Fang: Data curation; Formal analysis; Methodology. Wenbo Sun: Data curation; Formal analysis; Validation; Visualization. Shuyi Zhu: Data curation; Formal analysis; Validation. Qin Jiang: Project administration; Resources; Supervision. Xiaowei Wei: Funding acquisition; Project administration; Resources; Supervision; Writing—original draft. Yong Xu: Conceptualization; Project administration; Resources; Supervision; Writing—original draft; Writing—review & editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no potential conflict of interest.

Supporting information

Supporting Information

Wang, X. , Xu, F. , Kou, H. , Zheng, Y. , Yang, J. , Xu, Z. , Fang, Y. , Sun, W. , Zhu, S. , Jiang, Q. , Wei, X. , & Xu, Y. (2023). Stromal cell‐derived small extracellular vesicles enhance radioresistance of prostate cancer cells via interleukin‐8‐induced autophagy. Journal of Extracellular Vesicles, 12, e12342. 10.1002/jev2.12342