WERI-Rb1 human retinoblastoma cells (American Type Culture Collection) were cultured in RPMI-1640 (Gibco; Thermo Fisher Scientific, Inc.) with 10% foetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc.). Murine macrophages RAW264.7 (The Cell Bank of Type Culture Collection of the Chinese Academy of Sciences) were cultured in DMEM (Gibco; Thermo Fisher Scientific, Inc.) with 10% FBS. Bone marrow mesenchymal stem cells (BMSCs) were acquired from C57 mice and BMSCs from passages 4–6 were cultured in DMEM with 10% FBS for use (8×10⁴/ml). These cells were incubated in a humidified atmosphere with a mixture of 1% O₂, 5% CO₂ and 94% N₂ at 37°C.

BMSCs were isolated from the bone marrow of C57 BL/6 mice using previously described methods (20). Briefly, 4–6 week old C57 BL/6 mice were anesthetized by intraperitoneal injection of Nembutal (30 mg/kg; cat. no. P3761, Sigma-Aldrich; Merck KGaA) and then sacrificed by cervical dislocation, and their tibias and femurs were separated aseptically. DMEM was used to flush the marrow out with a sterile 25-guage needle and the extruded BM cells were filtered through a 70-µm cell strainer (BD Biosciences) and centrifuged at 1,000 × g for 5 min at 37°C. The BM cells were resuspended with DMEM containing 10% FBS and plated into flasks. The cells were treated with 0.25% trypsin/1 mM EDTA for 2 min until the cells reached 90% confluence. The culture medium was changed every 3 days. After 3–4 passages, isolated cells were used for experiments.

EXOs were purified from WERI-Rb1-derived conditioned medium (collected from 48 h cell cultures with RPMI-1640 and 10% exosome-free FBS) by ultracentrifugation, as outlined previously (21). The FBS was depleted of EXOs through ultracentrifugation at 110,000 × g overnight at 4°C. After 48 h, the conditioned media were centrifuged at 1,000 × g for 5 min, followed by 10,000 × g for 30 min, and then ultracentrifugation of the supernatants was performed at 100,000 × g for 90 min. All centrifugation was performed at 4°C. The EXOs were washed with phosphate-buffered saline (PBS), and then ultracentrifugation was carried out at 100,000 × g for 90 min at 4°C, following which, EXOs were resuspended in PBS. The size and number of EXOs were counted by Nano-Flow cytometry, according to a previous method (22), which is the imaging of individual fluorescently labelled EXOs passing through nanochannels in a pressure-driven flow (23). Lastly, the protein concentration was measured using a Bicinchoninic Acid Protein Assay Kit (Beyotime Institute of Biotechnology).

EXOs were viewed by transmission electron microscopy (FEI Tecnai Spirit G2; Thermo Fisher Scientific, Inc.), according to a previous method (21). Briefly, freshly isolated EXO pellets were transferred to a copper grid coated with carbon in a 30-µl drop of 1% glutaraldehyde for 5 min and then the grid was washed with distilled water. The grids were stained with 4% uranyl-acetate solution for 10 min at 37°C, and then treated with a 50-µl drop of methyl cellulose for 5 min on ice. Then, the grid was washed with distilled water and observed under transmission electron microscopy at 80 Kv.

For EXO labelling, EXOs were fluorescently labelled using the PKH26 membrane dye (Sigma-Aldrich; Merck KGaA), according to a previous method (24). Briefly, EXOs were resuspended into 1 ml Diluent C and 6 µl PKH26 dye was added for 5 min at room temperature. Then, the reaction was quenched by adding 2 ml 10% BSA (cat. no. 0332; Amresco, LLC) followed by ultracentrifugation at 100,000 × g for 2 h to remove excess dye.

For the co-culture experiments, BMSCs (4×10⁴) treated with PBS or EXOs were plated in 24-well plates. WERI-Rb1 cells (4×10⁴) were suspended in EXO-free medium and seeded into Transwell chambers with 0.4-µm pore size inserts (BD Biosciences). The viability of BMSCs and WERI-Rb1 cells were assessed using a CCK-8 assay after 24 h. BMSC RNA was extracted for RT-qPCR analysis.

The viabilities of WERI-Rb1 cells and RAW264.7 macrophages treated with EXOs or PBS were detected using a Cell Counting Kit-8 (CCK-8) assay (Dojindo Molecular Technologies, Inc.). Each group of cells was incubated with CCK-8 reagent for 3 h at 37°C. Finally, the absorbance was assessed at 450 nm and recorded.

Total RNA from WERI-Rb1 cells, RAW264.7 cells and BMSCs was extracted with TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Total RNA (1 µg) was subjected to reverse transcription using a PrimeScript™ RT Reagent kit (Takara Biotechnology Co., Ltd.), following the manufacturer's protocol. Total RNA from exosomes was extracted with TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) and Dr GenTLE™ Precipitation Carrier (Takara Biotechnology Co., Ltd.). Total RNA (1 µg) was subjected to reverse transcription using a Mir-X™ miRNA First-Strand Synthesis Kit (Takara Biotechnology Co., Ltd.), following the manufacturer's protocol. RT-qPCR was performed with a SYBR Prime Script RT-PCR Kit on a Roche 480 system (Roche Diagnostics) following the manufacturer's protocols. PCR conditions were as follows: 94°C for 5 min, followed by 40 cycles of 94°C for 30 sec, 62°C for 30 sec, and 72°C for 30 sec. The relative target gene expression was quantitated using the 2^−ΔΔCq method (25) and normalized to the endogenous expression of b-actin, GAPDH or miR-16. The primers for mRNA and miRNA detection are listed in Tables SI and SII, respectively.

Cells or tissue sections (tumour, spleen and liver) were fixed with 4% paraformaldehyde (10 min at 37°C), and then blocked with 1% BSA and 0.2% Triton X-100 for 30 min at 37°C. Samples were incubated with primary antibodies overnight at 4°C, including anti-CD49b for natural killer (NK) cells (1:100; cat. no. 14-5971-85; Invitrogen; Thermo Fisher Scientific, Inc.), anti-CD45 for leukocytes (1:100; cat. no. sc-1178; Santa Cruz Biotechnology, Inc.), anti-CD68 for macrophages (1:100; cat. no. BA3638; Wuhan Boster Biological Technology Co., Ltd.), anti-Ki-67 for proliferative cells (1:500; cat. no. MA5-14520; Thermo Fisher Scientific, Inc.), anti-Vimentin for invasive tumour cells (1:100; cat. no. sc-6260; Santa Cruz Biotechnology, Inc.). After washing with PBS, samples were stained with the appropriate Alexa Fluor-conjugated secondary antibody for 1 h at 37°C (Alexa Fluor 555 anti-rabbit lgG, 1:500, cat. no. 4413; Alexa Fluor 488 anti-rabbit lgG, 1:500, cat. no. 4412; Alexa Fluor 488 anti-mouse lgG, 1:500, cat. no. 4408; all purchased from Cell Signaling Technology, Inc.). Images were captured using fluorescence microscopy (Axio Imager Z1; Carl Zeiss AG; magnification, ×100).

Immunofluorescence staining for cells was quantified by analysing the fraction or number of positively-stained pixels per total field using ImageJ software (version 1.51; National Institutes of Health). This protocol was used in a previous study (26). Briefly, the images were binarized to black and white with a common threshold level such that white pixels represented positive cells and were quantified by histogram analysis using GraphPad Prism system 7 (GraphPad Software, Inc.). At least five random sections per mouse were used for quantification.

Whole proteins of cells and EXOs were extracted using a RIPA lysate kit (Beyotime Institute of Biotechnology). The protein concentration was determined with a BCA assay (Beyotime Institute of Biotechnology). Total protein (30 µg) from each sample was separated via 10% sodium dodecyl sulfate-polyacrylamide gel by electrophoresis, and subsequently transferred to a polyvinylidene difluoride membrane. The membranes were blocked with 5% BSA for 2 h at 37°C, then incubated with primary antibodies overnight at 4°C. The following primary antibodies were used: Anti-GAPDH (1:10,000; cat. no. 10494-1-AP; ProteinTech Group, Inc.), anti-Tubulin (1:1,000; cat. no. sc-5274; Santa Cruz Biotechnology, Inc.), anti-CD63 (1:1,000; cat. no. EXOAB-CD63A-1; Systems Biosciences, LLC), anti-tumour susceptibility gene 101 protein (1:1,000; cat. no. ab125011; Abcam), anti-CD9 (1:1,000; cat. no. 13403; Cell Signalling Technology, Inc.), anti-heat shock 70 kDa protein 1A (1:1,000; cat. no. 4873; Cell Signalling Technology, Inc.), anti-proliferating cell nuclear antigen (PCNA; 1:1,000; cat. no. 13110; Cell Signalling Technology, Inc.), anti-Bax (1:1,000; cat. no. 2772; Cell Signalling Technology, Inc.), anti-Bcl-2 (1:1,000; cat. no. A00040-1; Wuhan Boster Biological Technology, Ltd.), anti-Caspase-3 (1:1,000; cat. no. 9662; Cell Signalling Technology, Inc.), anti-C-X-C chemokine receptor type 4 (CXCR4; 1:1,000; cat. no. ab124824; Abcam), anti-MMP2 (1:1,000; cat. no. 10373-2-AP; ProteinTech Group, Inc.) and anti-thrombospondin-1 (TSP-1; 1:1,000; cat. no. ab1823; Abcam). Membranes were then incubated with HRP-conjugated anti-mouse IgG (1:3,000; cat. no. 7076s; Cell Signalling Technology, Inc.) or anti-rabbit IgG (1:10,000; cat. no. 7074s; Cell Signalling Technology, Inc.) for 1 h at 37°C, following which the blots were visualized with an enhanced chemiluminescence system (Thermo Fisher Scientific, Inc.). GAPDH or Tubulin were used as loading controls. The band intensity was semi-quantified by ImageJ software (version 1.51; National Institutes of Health).

A total of 20 female athymic nude mice (4–6 weeks old) were subcutaneously injected in the left submaxillary region with 300 µl WERI-Rb1 cells (1×10⁷) mixed with Matrigel (18 mg/ml, vol/vol, 1:1). One week after subcutaneous WERI-Rb1 cell injection, successfully transplanted mice (tumour volume, >200 mm³) were randomly divided into two groups (n=10) and subcutaneously injected with an equal volume of PBS or EXOs (75 µg) into the site of primary tumours every 3 days for 15 days. The tumour sizes of the mice were measured and the tumour volume was calculated as the length × width²/2 every 5 days. Mice were anesthetized by intraperitoneal injection of Nembutal (30 mg/kg) and then sacrificed by cervical dislocation 15 days after PBS or EXO injection. Tumours, spleens and livers were harvested and stored at −80°C. Sham surgery that did not include the injection of tumour cells, PBS and EXOs was performed on three female athymic nude mice (4–6 weeks old).

A total of 23 female nude mice (age, 4–6 weeks; weight, 16–18 g) and four C57 BL/6 mice (age, 4–6 weeks; weight, 16–20 g) were obtained from the Ophthalmic Animal Laboratory, Zhongshan Ophthalmic Center, Sun Yat-sen University (Guangzhou, China). All animal experiments adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research (27) and were approved and monitored by the Institutional Animal Care and Use Committee of Zhongshan Ophthalmic Center [approval no. SYXK (YUE) 2018-101, 2018-168, 2019-009]. The mice had free access to food and water and were maintained under a 12-h light/dark cycle in an air-conditioned room (16-26°C and 40–70%).

Data are presented as the mean ± standard deviation, and the differences between mean values were evaluated using a Student's two-tailed t-test (for two groups) or analysis of variance followed by Tukey's post hoc (for multiple group comparisons). P<0.05 was considered to indicate a statistically significant difference.

WERI-Rb1 cells grew in loose grape-like clusters and were cultured in EXO-free media (Fig. 1A). EXOs were isolated from cells, as described in the materials and methods section, and observed under a transmission electron microscopy. The diameter of vesicles was ~30–100 nm (Fig. 1B). The NanoFCM tracking analysis also indicated that the size distribution and number of exosomes derived from WERI-Rb1 cells, ranged from 50 to 150 nm with a mean size of ~77 nm (Fig. 1C). Additionally, known exosome putative markers, including HSP70, TSG101, CD9 and CD63, were identified by western blotting (Fig. 1D). To determine whether EXOs could affect the growth of retinoblastoma cells, WERI-Rb1 cells were incubated with EXOs labelled with PKH26 (red). As presented in Fig. 1E, EXOs were taken up by WERI-Rb1 cells at 24 h after incubation. The CCK-8 assay indicated that the viability of WERI-Rb1 cells treated with EXOs derived from the same cells was significantly increased compared with that of the control (PBS, 0.122±0.002; EXO, 0.145±0.005; *P<0.05; Fig. 1F). However, the mRNA expression levels of VEGF, PCNA and CXCR4, which can be used to predict cancer cell malignancy (28–30), were not affected by EXOs [no significance (ns); Fig. 1G]. These data suggested that EXOs derived from WERI-Rb1 cells have no significant influence on malignancy in vitro.

To reveal the effect of EXOs on immune cells, RAW264.7 cells, which are well-characterized with regard to their macrophage-specific immune functions (31), were incubated with EXOs from WERI-Rb1 cells labelled with PKH26 (red). At 6 h after incubation, EXOs could be taken up by RAW264.7 cells. PKH26 was partially colocalized with CD68 (green, a specific marker of macrophages; Fig. 2A). Moreover, the cells changed from round to spindle-shaped with ramified morphology (Fig. 2B), which indicated that macrophages were activated. However, EXOs significantly increased the levels of IL-6 and MCP-1 (relative fold-change: IL-6, 2.30±1.11-fold; MCP-1, 1.43±0.43-fold; *P<0.05). The increase in IL-6 and MCP-1 expression implied that the antitumour activity of macrophages was inhibited by EXOs, which was consistent with a previous study (6). There was no significant difference in TNF-α expression between the PBS and EXO groups (relative fold-change: 1.16±0.28-fold; ns; Fig. 2C). It has been speculated that TNF-α expression does not change in the early response of RAW 264.7 cells to EXOs (24 h). Accordingly, treatment with EXOs also resulted in decreased proliferation of RAW264.7 cells (PBS, 0.949±0.271; EXO, 0.303±0.124; *P<0.05; Fig. 2D).

To assess the effect of EXOs on the bioactivity of BMSCs and WERI-Rb1 cells, a co-culture system was used to mimic retinoblastoma physiological conditions. The primary BMSCs were incubated with PBS or EXOs and co-cultured with WERI-Rb1 cells for 24 h (Fig. 2E). As shown in Fig. 2F, EXOs could significantly increase the expression levels of IL-6 and MCP-1 in BMSCs (relative fold-change: IL-6, 7.54±1.50-fold; MCP-1, 7.62±1.39-fold; *P<0.05), which could promote tumour growth and metastasis (32,33). Similarly, there was no significant difference in TNF-α expression between the PBS and EXO groups in BMSCs (relative fold-change: 0.66±0.26-fold; ns). Moreover, the viability of WERI-Rb1 cells was significantly increased (PBS, 0.14±0.03; EXO, 0.21±0.02; *P<0.05) in the co-culture system, whereas the viability of BMSCs was not affected by EXOs (PBS, 0.42±0.05; EXO, 0.46±0.02; ns) (Fig. 2G).

Taken together, these findings demonstrated that EXOs derived from WERI-Rb1 cells could inhibit the antitumour activity of macrophages and induce the BMSCs to promote retinoblastoma cell growth in vitro.

To confirm the in vitro results, a xenotransplantation model was established and then mice were regularly injected with EXOs or PBS around the tumour. EXOs in tumour tissues were observed after 2 days and were markedly increased in the loose tumour tissue at day 10. It was observed that EXOs were mainly located in loose tissue outside the tumour, and few were detected in the dense tissue within the tumour (Fig. 3A). However, quantification of tumour volume growth showed that EXO treatment significantly increased retinoblastoma tumour volume 10 days after transplantation (PBS, 66±172 mm³; EXO, 259±224 mm³; *P<0.05; Fig. 3B and C). Moreover, the protein levels of PCNA were significantly elevated by EXOs in the tumour (PBS, 0.77±0.18; EXO, 1.08±0.31; *P<0.05; Fig. 3D). Furthermore, the protein levels of apoptosis-related Bcl-2, Bax and caspase-3 were detected. As presented in Fig. 3E, the ratio of Bcl-2 to Bax was significantly increased (PBS, 0.58±0.23; EXO, 1.44±0.87; *P<0.05), whereas the relative expression of cleaved caspase-3 was decreased (PBS, 0.28±0.18; EXO, 0.17±0.14; *P<0.05) in the EXO group. In summary, EXOs derived from WERI-Rb1 cells could inhibit tumour apoptosis and promote tumour growth in vivo.

Ki-67 is widely used to measure tumour proliferation and assess the prognosis of patients with cancer (34). Vimentin, an intermediate filament protein, is often used as a marker of invasive tumour cells due to its expression during activation of the epithelial-mesenchymal transition (35). Therefore, tissue sections were double stained with antibodies against Ki-67 and Vimentin. As shown in Fig. 4A and B, the relative Vimentin-positive area (PBS, 11.35±2.43%; EXO, 21.04±3.58%; *P<0.05) and Ki-67-positive rate in cells were significantly increased (PBS, 45.41±9.39; EXO, 69.20±2.62; *P<0.05) in tumour tissues injected with EXOs. Moreover, the protein levels of CXCR4 and MMP2, which play a crucial role in malignancy and metastasis during tumour progression (36,37), were significantly increased (CXCR4: PBS, 0.225±0.088; EXO, 0.567±0.238; MMP2: PBS, 0.330±0.189; EXO, 0.802±0.433; *P<0.05; Fig. 4C and D).

Furthermore, TAMs, a significant subset of tumour-infiltrating immune cells, play an important role in tumour growth and metastasis (38,39). Therefore, TAMs were evaluated in whole tumour sections using staining with an anti-CD68 antibody (Fig. 4E). Accordingly, the CD68-positive area was significantly increased in tumour tissues treated with EXOs compared with that in control tissue (PBS, 1.18±0.41%; EXO, 4.50±3.53%; *P<0.05; Fig. 4F). NK cells labelled with CD49b (red) surrounded tumour cells and were decreased in the EXO-treated group (PBS, 14.94±3.77; EXO, 3.66±2.29; ***P<0.001; Fig. 4G and H). Therefore, these results also partially supported the aforementioned results that demonstrated that exosomes derived from WERI-Rb1 cells promoted tumour growth and malignancy.

Although athymic nude mice with immunodeficiency were used in the present study, the mice still exhibited intact innate and adaptive immunity. Thus, the spleen was analysed to determine whether EXOs affect immune processes. Of note, it was found that EXOs could diffuse into the spleen after 2 days and continue to accumulate at day 10 (Fig. 5A and B). EXOs were also observed in CD68-positive macrophages in spleen tissues (Fig. 5C).

Additionally, NK cells can limit tumourigenesis in the absence of functioning T-lymphocytes (40,41). Therefore, spleen tissues were double stained with antibodies against Ki-67 and CD45 to determine the number and proliferative activity of leukocytes (Fig. 5D). CD45-positive leukocytes, which include granulocytes, monocytes and lymphocytes, control the innate immune response (42). The present data showed that EXOs significantly decreased the number of leukocytes in the spleen (PBS, 109.30±52.11; EXO, 37.54±21.42; *P<0.05; Fig. 5E). More specifically, EXOs also significantly decreased the ratio of cells co-expressing Ki-67 and CD45 (white triangle), which indicated that compared with the control, EXOs inhibited leukocyte proliferation in the spleen (PBS, 28.87±10.85; EXO, 8.47±5.39; **P<0.01; Fig. 5F). Similarly, immunofluorescence staining of CD49b, a pan-marker of NK cells (43), indicated that NK cells were significantly decreased in the spleens of the EXO-injected groups compared with the sham operation and PBS-injected groups (sham, 41.28±6.24; PBS, 37.81±18.08; EXO, 14.84±5.84; *P<0.05, ***P<0.001; Fig. 5G and H). Collectively, these data demonstrated that EXOs inhibited the innate immune response in vivo.

To explore the effects of EXOs on retinoblastoma metastasis, the liver, which was the closest organ to the subcutaneous tumour, was analysed. As shown in Fig. 6A, EXOs were observed in the cells around the liver sinusoids 2 days following the EXO injection and only appeared in the cells near the liver sinusoids. At day 10, EXOs (red) appeared in distant liver tissues and were mainly engulfed by Kupffer cells (CD68-positive), demonstrating that EXOs could spread throughout liver tissues and accumulate in macrophages (white arrows; Fig. 6B). Additionally, macroscopic images of liver metastatic nodules shown in Fig. 6C indicated that EXO injection led to more metastatic lesions (white arrowheads, metastatic liver nodules). Three livers from the EXO-injection group were cirrhotic, which indicated liver deterioration (green arrowheads, liver cirrhosis). Then, the excised livers were stained with Ki-67. It was found that livers from the EXO-injected group exhibited increased Ki-67-positive areas, which was consistent with metastatic nodules in the liver tissues (white arrowheads; PBS, 0.405±0.353%; EXO, 3.926±1.533%; **P<0.01; Fig. 6D and E). Thus, EXOs derived from WERI-Rb1 cells may exhibit a promoting effect on tumour metastasis.

Since EXOs derived from WERI-Rb1 cells significantly affected tumour malignancy, innate immunity and metastasis as aforementioned, the expression of a selection of miRNAs and proteins involved in tumour deterioration in EXOs were analysed via RT-qPCR and western blotting. As shown in Fig. 7A, miR-92a, miR-20a, miR-129 and miR-17 were detected. Moreover, the chemokine receptor CXCR4 was not only expressed in WERI-Rb1 cells but was also detected in EXOs (Fig. 7B). TSP-1, a matricellular protein, plays multiple roles in tumour development and was not observed in WERI-Rb1 cells, which was consistent with our previous study (44). However, of note, TSP-1 was detected in EXOs. CXCR4 and TSP-1 are involved in proliferation and TAM recruitment (45–51). Thus, the contents of EXOs may play a key role in EXO-mediated tumour progression.