The DLBCL cell lines U2932 (RRID: CVCL_1896) and OCI-LY8 (RRID: CVCL_8803) were purchased from Nanjing Cobioer Biotechnology Co., Ltd., the human gastric mucosal epithelial cell line GES-1 (RRID: CVCL_EQ22) was kindly provided by the Institute of Oncology, Nanhua University (Hengyang, China), and the human T cell line H9 (RRID: CVCL_1240) was purchased from Shanghai Yiyan Biotechnology Co., Ltd. U2932, OCI-LY8 and H9 cells were cultured in RPMI 1640 medium (Thermo Fisher Scientific, Inc.) with 10% (v/v) heat inactivated fetal bovine serum (FBS; Thermo Fisher Scientific, Inc.), 100 units/ml penicillin (Thermo Fisher Scientific, Inc.) and 100 µg/ml streptomycin (Thermo Fisher Scientific, Inc.). H9 cells were cultured in 75 cm² tissue culture flasks at 37°C in an incubator with 5% CO₂ and 95% humidity, and the culture medium was refreshed every 2–4 days. H9 cells were centrifuged at 125 × g at room temperature for 5–10 min and resuspended in fresh medium at 5×10⁵ cells/ml to remove cell debris and replace the medium. For good growth states, H9 cells were maintained at cell concentrations between 5×10⁵ and 2×10⁶ cells/ml. Passage of U2932 and OCI-LY8 cells was performed when cells reached 80–90% confluence. U2932 and OCI-LY8 cells were sequentially centrifuged at 125 × g at room temperature for 5 min, resuspended in 2 ml fresh medium, seeded into a new 25 cm² cell culture bottle with 8 ml fresh culture medium. The human normal gastric mucosal GES-1 cell line was cultured in Dulbecco's modified Eagle's medium (Thermo Fisher Scientific, Inc.) with 10% (v/v) FBS, 100 U/ml penicillin (Thermo Fisher Scientific, Inc.) and 100 µg/ml streptomycin (Thermo Fisher Scientific, Inc.). The FBS used for cultivation of all types of cells was ultracentrifuged at 100,000 × g and 4°C for 10 h to remove exosomes. All kinds of cells were cultured at 37°C in a cell incubator with 5% CO₂.

The total cellular RNA isolated using TRIzol^® regent (Thermo Fisher Scientific, Inc.) from U2932 cells was used for synthesis of complementary DNA using the PrimeScript™ RT reagent Kit with gDNA Eraser (Takara Biotechnology Co., Ltd.). Sequences used for plasmid construction were obtained using a high-fidelity PCR kit KOD-Plus-Neo (Toyobo Life Science) and cloned into the p3×Flag-CMV-14 vector (MilliporeSigma). The primers used for plasmid construction were as follows: forward, 5′-GGGGTACCATGAGGATATTTGCTGTCTTT-3′ and reverse, 5′-GCTCTAGACGTCTCCTCCAAATGTGTAGT-3′. Gene silencing plasmids for PD-L1 were constructed using the pRNAT-U6.1/neo vector (GenScript) according to the manufacturer's protocols. The target sequence used for constructing the gene silencing plasmid shRNA-PD-L1 was 5′-GCATTTGCTGAACGCATTT-3′. The negative control (target sequence: 5′-ACTACCGTTGTTATAGGTG-3′) was constructed previously (12). All plasmids were amplified using Escherichia coli DH5α competent cells and were sequenced by Sangon Biotech Co., Ltd to confirm the correct construction of the plasmid.

Cell transfection of plasmids was performed according to the manufacturer's protocols. Briefly, a total of 1.5×10⁶ cells at the logarithmic growth stage were seeded into each well of a 12-well cell culture plate. A total of 1 µg plasmids and 4 µl Lipofectamine^® 2000 (Thermo Fisher Scientific, Inc.) were mixed in 100 µl of Opti-MEM^TM medium (Thermo Fisher Scientific, Inc.), separately. After incubation for 5 min at room temperature, plasmids and Lipofectamine^® 2000 were mixed together and incubated at room temperature for 20 min. Then, 200 µl of the mixture was added to each well of the 12-well cell culture plate. After transfection at 37°C for 12 h, the culture medium was replaced with fresh medium. The cells were used for further experiments 24–48 h after transfection.

Exosomes in the supernatants of cultured cells were extracted by ultracentrifugation according to previously reported protocols (28). Briefly, the supernatants of cultured cells were centrifuged at 2,000 × g and 4°C for 20 min and then transferred into a centrifuge tube. The supernatants were then centrifuged at 10,000 × g and 4°C for 30 min and then transferred into new ultracentrifuge tubes. The supernatants were then ultracentrifuged at 100,000 × g and 4°C for 70 min, and the remaining supernatants were discarded. The pellets were resuspended in PBS and ultracentrifuged at 100,000 × g and 4°C for 70 min. The pellets were considered to be the extracted exosomes, which were resuspended in PBS or RIPA lysis buffer (Thermo Fisher Scientific, Inc.) with proteinase inhibitors. Exosomes in the plasma of GDLBCL patients and healthy individuals were extracted using the Exoquick TC kit (System Biosciences, LLC) according to the manufacturer's protocols. The extracted exosomes were identified by nanoparticle tracking analysis (NTA) and transmission electron microscopy (TEM) as previously reported (29). Protein expression levels of specific markers of exosomes, CD9 and CD63, were assessed using immunoblotting.

The DLBCL cells were lysed using RIPA lysis buffer (Thermo Fisher Scientific, Inc.) with proteinase inhibitors. After incubation on ice for 15–30 min, lysates were centrifuged at 16,100 × g and 4°C for 15 min. After determination of the protein concentration by the BCA method, 40 µg total protein was sequentially run on 10% SDS-PAGE gels and transferred onto PVDF membranes which were blocked with 5% skim milk at room temperature for 1 h. Membranes were incubated with the primary antibodies at 4°C overnight, washed using TBST buffer with 1% Tween-20, and incubated with the secondary antibodies at room temperature for 2 h. Protein bands were detected using the SuperSignal^® West Pico Chemiluminescent Substrate kit (Thermo Fisher Scientific, Inc.). β-actin was used as the internal control for total protein, and CD9 and CD63 were used as the internal controls for exosomes. Densitometric analysis of protein bands was performed using software Image Lab (version 5.2; Bio-Rad Laboratories, Inc.). The antibodies used in the present study were as follows: Rabbit anti-PD-L1 (1:1,000, clone E1L3N, cat. no. 13684, Cell Signaling Technology, Inc.), mouse anti-β-actin (1:5,000, clone AC-15, A5441, Sigma-Aldrich; Merck KGaA), mouse anti-Flag (1:5,000, clone M2, F3165, MilliporeSigma), mouse anti-CD9 (1:200, clone C-4, sc-13118, Santa Cruz Biotechnology, Inc.), mouse anti-CD63 (1:200, clone MX-49.129.5, cat. no. sc-5275, Santa Cruz Biotechnology, Inc.), goat anti-mouse IgG (HRP-linked, 1:5,000, AP124P, MilliporeSigma), goat anti-rabbit IgG (HRP-linked, 1:5,000, AP132P, MilliporeSigma).

Briefly, an average of 1×10⁴ stably transfected DLBCL cells in 100 µl of culture medium were seeded per well of a 96-well cell culture plate. After culturing at 37°C for 24 h, 10 µl of 5 mg/ml MTT solution was added to the culture medium and incubated at 37°C for 4 h. Then, the culture medium was carefully removed and replaced with 150 µl of DMSO. The OD value at 490 nm was quantified using a Synergy HTX microplate reader (BioTek Instruments, Inc.). Experiments were repeated three times.

A total of 12 NOD-SCID mice (age, 6 weeks; average weight, 20 g; Hunan Slack Jingda Experimental Animal Co., Ltd.) were used to perform animal experiments. Mice were fed under standard conditions (25°C and 50% humidity) with free access to sterile feed and water in a pathogen-free environment with a 12 h light/dark cycle at the animal care facility of Hunan Cancer Hospital (Changsha, Hunan, China). The mice were randomly assigned to two groups with six mice per group. An average of 5×10⁶ U2932 cells in 200 µl of sterile PBS were subcutaneously injected into all mice. Seven days after cell injection, 100 µg of exosomes derived from U2932 cells stably transfected with the p3×Flag vector or p3×Flag-PD-L1 in 100 µl of sterile PBS were administered to the mice every four days through the tail vein. Mice were sacrificed when they experienced a sharp decrease in activity, water and diet intake, if the tumors formed under the skin of mice were assessed to be about to reach 15 mm in diameter in any dimension or if a total of 30 days after cell injection was reached. The mice were sacrificed by cervical dislocation immediately after isoflurane anesthesia (induction, 3%; maintenance, 1%). The formed tumors were measured every two days and recorded for further analysis.

All tissue specimens used in the present study were obtained from Hunan Cancer Hospital with the informed consent of patients and the present study was approved by the institutional review boards of Hunan Cancer Hospital (approval no. 2021-012) and was performed in accordance with the Declaration of Helsinki. The blood samples of 26 GDLBCL patients were collected from Hunan Cancer Hospital from January 2017 to December 2020. The inclusion criteria used were as follows: i) the stomach was the primary site, which may have been accompanied by lymph node metastasis in the gastric drainage area; and ii) the pathological diagnosis was DLBCL. Samples that failed to meet both of the above inclusion criteria were excluded. A total of 10 males and 16 females, aged 30–69 years old with a median age of 51.2 years were included. The patients were not treated with antineoplastic therapy before samples were taken. The Lugano stage (2016) (30) of patients was stages I–II in 15 cases and stages III–IV in 11 cases. The international prognostic index (IPI) score (31) of the lymphomas was 0–2 in 16 cases and 3–5 in 10 cases. The histological classification was 14 cases in the non-germinal center B-cell like lymphoma (non-GCB) subtype and 12 cases in the GCB subtype. A total of 22 healthy individuals who volunteered in the same period were selected as the normal control group, which included 10 males and 12 females (median age, 48.5 years; age range, 26–62 years).

Tissue sections used for IHC were assessed by two pathologists at Hunan Cancer Hospital (Changsha, Hunan, China), and IHC staining was performed as previously reported (12). Briefly, fresh tissues collected immediately after surgery were routinely fixed using 10% formalin solution at room temperature for 24 h, embedded in paraffin, followed by slicing into 5 µm sections. Then, tissue sections were dewaxed in xylene at room temperature and rehydrated in graded alcohols, and tissue antigen retrieval was performed in boiled 1 µM sodium citrate solution (pH, 6.0) at 100°C for 2 min. After blocking with normal goat serum (Thermo Fisher Scientific, Inc.) at room temperature for 1 h, slices were sequentially incubated with the primary antibodies at 4°C overnight and secondary antibodies for 2 h at room temperature, stained with DAB solution at room temperature for 5 min, and counterstained with hematoxylin at room temperature for 1 min. The antibodies used for IHC staining were as follows: Rabbit anti-PD-1 (1:200, clone D4W2J, cat. no. 86163S, Cell Signaling Technologies, Inc.), rabbit anti-PD-L1 (1:500, 17952-1-AP, Proteintech Group, Inc.), mouse anti-CD8 (1:5,000, clone 1G2B10, cat. no. 66868-1-Ig, Proteintech Group, Inc.), rabbit anti-CD5 (1:800, clone E8X3S, cat. no. 39300S, Cell Signaling Technologies, Inc.), rabbit anti-CD10 (1:500, clone E5P7S, cat. no. 65534S, Cell Signaling Technologies, Inc.), rabbit anti-CD20 (1:200, clone E7B7T, cat. no. 48750S, Cell Signaling Technologies, Inc.), rabbit anti-CD79a (1:250, clone D1X5C, cat. no. 13333S, Cell Signaling Technologies, Inc.), mouse anti-Ki67 (1:2,000, clone 8D5, cat. no. 9449S, Cell Signaling Technologies, Inc.), goat anti-mouse IgG (HRP-linked, 1:2,000, cat. no. AP124P, MilliporeSigma), and goat anti-rabbit IgG (HRP-linked, 1:2,000, cat. no. AP132P, MilliporeSigma).

Tissue slices were viewed under an inverted microscope, and representative images were presented in the figures. Three fields of view per section were analyzed. Positivity for CD5, CD10, CD8, PD-L1 and PD-1 was defined as ≥5% positively stained cells, and positivity for CD20, CD79a and Ki67 was defined as ≥50% positively stained cells. One-fifth of the cases were scored by two observers to assess reproducibility. For cases to be considered suitable for evaluation, ≥25% area of a tissue slice had to be available for morphologic analysis following staining and at least one positively stained tumor-infiltrating macrophage was required as a positive internal control. Analysis of IHC staining of CD3, CD5, CD8, CD10 and CD20 was performed according to the manufacturers' protocols and as previously reported (32).

Data analysis was performed using SPSS version 22.0 (IBM, Corp.). Statistical graphs were drawn using GraphPad Prism 5.0 (GraphPad Software; Dotmatics). Unpaired, two-sided Student's t-test was used to assess the statistical significance. The relationship between the PD-L1 level in plasma exosomes and the clinicopathological characteristics of GDLBCL patients was determined using the χ² or Fisher's exact test. Data were presented as the mean ± SD from three independent biological replicates. P<0.05 was considered to indicate a statistically significant difference.

To assess the regulatory role of exosomal PD-L1 in the immune microenvironment of GDLBCL, U2932 and OCI-LY8 DLBCL cells were chosen as experimental cell models and the supernatants of the cell culture medium were collected to extract exosomes by ultracentrifugation. The protein markers CD9 and CD63 detected using immunoblotting indicated the successful extraction of exosomes (Fig. 1A). NTA demonstrated that the particle size of exosomes in the supernatants of cultured cells presented a normal distribution and was mainly concentrated between 70 and 120 nm (Fig. 1B). Interestingly, compared with the protein expression level of exosomal PD-L1 derived from human epithelial gastric mucosal GES-1 cells, the protein expression level of PD-L1 in exosomes derived from DLBCL cells U2932 and OCI-LY8 was much higher (Fig. 1A). Wit was hypothesized that the high protein expression level of exosomal PD-L1 might facilitate the initiation and development of GDLBCL.

Next, Flag-tagged PD-L1 was overexpressed or PD-L1 expression was silenced using shRNA gene-silencing plasmids in U2932 and OCI-LY8 cells. Assessment of the protein content using immunoblotting, demonstrated that Flag-tagged PD-L1 appeared in exosomes derived from the supernatants of U2932 and OCI-LY8 cells (Fig. 1C). Moreover, the protein expression level of exosomal PD-L1 was markedly decreased with the inhibition of cell intrinsic PD-L1 (Fig. 1D). These findings indicated that the protein expression level of exosomal PD-L1 was consistent with the protein expression level of cell-intrinsic PD-L1.

To further illustrate the regulatory role of exosomal PD-L1, tumor xenograft and tail vein injection experiments were performed in NOD-SCID mice. One week after injecting wild-type U2932 cells into mice, exosomes derived from the supernatants of U2932 cells overexpressing PD-L1 and the control vector were administered into mice through tail vein injection. Four weeks after cell injection, significantly larger and heavier tumors were observed in the mice injected with exosomes derived from PD-L1-overexpressing cells compared with that of the control (Fig. 1E). These results indicated that, exosomal PD-L1 contributed to tumor growth in vivo.

As exosomal PD-L1 contributes to the immune evasion of cancer cells, it was hypothesized that exosomal PD-L1 might influence T lymphocytes in the immune microenvironment. Therefore, the influence of exosomal PD-L1 from the supernatants of cultured DLBCL cells on the proliferation of H9 human T lymphocytes was assessed. In MTT experiments, exosomes with PD-L1 overexpression or PD-L1 silencing as well as those of controls derived from U2932 and OCI-LY8 cells were added to the culture medium of H9 cells. The proliferation of H9 cells was significantly inhibited by treatment with exosomes with PD-L1 overexpression compared with the control and was significantly promoted by treatment with exosomes with PD-L1 silencing compared with the control (Fig. 2). These results demonstrated that PD-L1 in exosomes derived from DLBCL cells inhibited the proliferation of T lymphocytes. Therefore, exosomal PD-L1 originating from DLBCL cells may interact with the surface PD-1 of T lymphocytes, resulting in immune suppression.

Exosomes from the plasma of GDLBCL patients and healthy individuals were obtained for analysis. The detection of CD9 and CD63 proteins indicated the successful extraction of plasma exosomes (Fig. 3A). The exosomes extracted from the plasma were also assessed using TEM (Fig. 3B). Interestingly, the protein expression level of exosomal PD-L1 in three representative GDLBCL patients was markedly higher than that in healthy individuals (Fig. 3C). The statistical analysis demonstrated significant upregulation of exosomal PD-L1 in the plasma of GDLBCL patients (Fig. 3D). These data indicated that exosomes with high levels of PD-L1 may participate in the occurrence and development of GDLBCL.

The correlation between the protein expression level of exosomal PD-L1 and clinicopathological features of GDLBCL was further analyzed, including gender, age, Lugano stage, IPI score and pathological subtypes of lymphoma (Table I). Although no significant relationship was demonstrated between the protein expression level of exosomal PD-L1 in plasma and gender or age, the protein expression level of exosomal PD-L1 was demonstrated to be significantly, positively related with the IPI score, non-GCB pathological type or Lugano stage of GDLBCL. These results indicated that a high level of exosomal PD-L1 may lead to malignant transformation and poor prognosis of GDLBCL.

To evaluate the relationship between exosomal PD-L1 in the plasma and the immune microenvironment of GDLBCL, the protein expression levels of CD20, CD79a, CD5, CD10, Ki67, CD8, PD-1 and PD-L1 were assessed using IHC staining in a series of consecutive slices of GDLBCL tissue specimens (Fig. 4A and B). The relationship between the protein expression level of PD-L1 in plasma exosomes and the immune microenvironment was evaluated (Table II). The protein expression levels of PD-L1 and CD8 in tissue specimens were significantly and positively related to the upregulation of PD-L1 expression in plasma exosomes (P=0.0004 and P=0.0183, respectively; Table II). Exosomal PD-L1 from gastric cancer cells interacts with the surface PD-1 on CD8⁺ T cells, which leads to immune evasion of cancer cells (21). Therefore, a high expression level of exosomal PD-L1 in the plasma may reflect the immunosuppressive microenvironment of GDLBCL. There was no significant relationship between the protein expression level of PD-L1 in plasma exosomes and that of CD20, CD79a, CD5, CD10, Ki67 or PD-1 in GDLBCL tissue specimens (Table II).