High-throughput screening identifies ibuprofen as an sEV PD-L1 inhibitor for synergistic cancer immunotherapy

Abstract

Programmed death-ligand 1 (PD-L1) on tumor-derived small extracellular vesicles (sEVs) limits therapeutic effectiveness by interacting with the PD-1 receptor on host immune cells. Targeting the secretion of sEV PD-L1 has emerged as a promising strategy to enhance immunotherapy. However, the lack of small-molecule inhibitors poses a challenge for clinical translation. In this study, we developed a target and phenotype dual-driven high-throughput screening strategy that combined virtual screening with nanoflow-based experimental verification. We identified ibuprofen (IBP) as a novel inhibitor that effectively targeted sEV PD-L1 secretion. IBP disrupted the biogenesis and secretion of PD-L1⁺ sEVs in tumor cells by physically interacting with a critical regulator of sEV biogenesis, hepatocyte growth factor-regulated tyrosine kinase substrate. Notably, the mechanism of action of IBP is distinct from its commonly known targets, cyclooxygenases. Administration of IBP stimulated antitumor immunity and enhanced the efficacy of anti-PD-1 therapy in melanoma and oral squamous cell carcinoma mouse models. To address potential adverse effects, we further developed an IBP gel for topical application, which demonstrated remarkable therapeutic efficacy when combined with anti-PD-1 treatment. The discovery of this specific small inhibitor provides a promising avenue for establishing durable, systemic antitumor immunity.

Keywords: drug repurposing; high-throughput screening; immunotherapy; sEV PD-L1; sEV inhibitors; synergistic cancer therapy; targeted therapy; tumor-derived small extracellular vesicles.

Graphical abstract

Figure 1

High-throughput screening for candidate compounds targeting HRS (A) Schematic diagram showing the TAP-HTS strategy. (B) We selected 434 compounds with Ki ≤200 μM (red) from the drug libraries. Compounds marked in blue are non-inhibitory. (C) We selected 258 candidate compounds (green) after safety evaluation by the PrOCTOR assays. (D) Schematic illustration depicting the validation of candidate compounds based on the NanoScreen assay. The proportion of PD-L1⁺ sEVs in the culture supernatant of CAL27 PD-L1-EGFP cells was quantified using a nanoflow analyzer. The WGA staining of sEVs and the corresponding gating strategy are available in Figure S1. (E) Viability analysis of the cells with indicated compounds treatment by CCK-8 assays (n = 2). (F) Quantification of secreted PD-L1⁺ sEVs in the cell supernatant by nanoflow assays (n = 2). Both CCK-8 and nanoflow assays were performed in two independent biological replicates. The average of the two results was used for analysis. (G) Seven compound candidates were determined by the combined analysis of inhibitory efficacy and cell cytotoxicity. Detailed results for all screened compounds are provided in Table S2. (H) Nanoflow-based experimental verification of seven compound candidates (n = 4). (I) A list of the structural formulas of the selected candidates. Data are represented as mean ± SEM. The significance of the results for each compound compared to the vehicle was determined using a two-tailed unpaired t test.

Figure 2

IBP inhibits the secretion of sEV PD-L1 from tumor cells (A) A representative TEM image of CAL27-derived sEVs. Scale bar, 100 nm. (B) The size distribution of sEVs from tumor cells with or without IBP treatment by NTA analysis (n = 3). (C) The secretion curve of sEVs derived from B16F10 cells with or without IBP treatment (n = 3). (D) Quantification of average protein content of single sEV by the BCA assays and NTA analysis (n = 3). (E–G) Western blot analysis of various proteins in sEVs and WCLs from CAL27 (E), H1264 (F), and B16F10 (G) cells with or without IBP treatment (n = 3). The particle concentration of sEV samples was determined by NTA. Subsequently, 5 × 10⁹ sEVs were loaded into each lane for western blot analysis. The relative sEV PD-L1 intensity was quantified by normalizing to the grayscale value of the control group in each independent experiment. The equal protein amount of WCL was loaded in each lane. GAPDH was used as loading control for WCL analysis. Data are represented as mean ± SEM. Significance was determined using a two-tailed unpaired t test.

Figure 3

IBP inhibits the secretion of sEV PD-L1 in a COX-independent manner (A) Western blot analysis of HRS, COX1, and COX2 in the indicated tumor cell lines (n = 3). (B) Images of RT-PCR results showing the expression of COX1 and COX2 in the indicated tumor cell lines (n = 3). (C) Western blot analysis of COX1 or COX2 in CAL27 COX1 KO and CAL27 COX2 KO cells (n = 3). (D) The secretion curve of sEVs derived from CAL27 COX1 KO and CAL27 COX2 KO cells (n = 3). (E) Western blot analysis of PD-L1 in sEVs and WCL from the indicated cells with or without IBP (100 μM) treatment (n = 5). (F) Western blot analysis of PD-L1 in sEVs and WCL from the B16F10 parental or HRS KO cells with or without IBP (100 μM) treatment (n = 5). The equal protein amount of cell lysate or equal number of sEVs was loaded in each lane. We loaded 5 × 10⁹ sEVs into each lane for western blot analysis. The relative sEV PD-L1 intensity was quantified by normalizing to the grayscale value of the control group in each independent experiment. The equal protein amount of WCL was loaded in each lane. GAPDH was used as loading control for WCL analysis. Data are represented as mean ± SEM. Significance was determined using a two-tailed unpaired t test.

Figure 4

IBP inhibits HRS-mediated protein cargoes sorting into TEVs (A) The calculated binding mode between IBP and HRS domains at the active site pockets. The IBP is represented by blue sticks. (B) Pull-down assays showing the interaction between HRS and IBP using the CAL27 WCLs. Streptavidin-conjugated beads were used to pull down the biotinylated IBP and its interacting proteins. WT-IBP was used as a negative control (n = 3). (C) CETSA of HRS in cell lysates of CAL27 cells with or without IBP treatment (500 μM) (n = 3). (D) Representative confocal immunofluorescent images showing the colocalization of HRS (green) with the endosomal markers Rab5 or Rab7 (red) in B16F10 cells with or without IBP treatment (200 μM). Hoechst staining shows the nucleus in blue. Scale bars, 10 μm (n = 3). (E) A total of nine areas from three biological replicate immunofluorescence samples were selected. Quantitative co-localization analysis was performed using Pearson correlation. Data are represented as mean ± SEM. Significance was determined using a two-sided unpaired Mann-Whitney test (Rab5) or a two-sided unpaired t test (Rab7). (F) The sEVs for MS analysis were purified from the culture supernatants of B16F10 cells with or without IBP treatment (200 μM). Pathway analysis shows the top upregulated (red) and downregulated (blue) clusters of the sEV proteins. The fold change (FC) of each detected protein is shown (FC cutoff at 1.2). (G and H) Enriched Gene Ontology Cellular Compartment (GOCC), Reactome, and Kyoto Encyclopedia of Genes and Genomes (KEGG) categories in the cohort of upregulated (red) or downregulated (blue) sEV proteins.

Figure 5

IBP stimulates anti-tumor immunity in the melanoma mouse model (A) Therapeutic schedule for the mice inoculated with B16F10 WT or HRS KO cells. (B) The growth curves of B16F10 WT tumors (n = 5) and B16F10 HRS KO tumors (n = 6) in C57BL/6 mice treated with IBP or DMSO. (C) Quantification of average protein content of single TEV by the BCA assays and NTA analysis (n = 5). (D) Western blot analysis of PD-L1 and HRS in the sEVs derived from B16F10 WT tumor tissues. Equal particle number of sEVs (5 × 10⁹) was loaded in each lane (n = 5). (E) Western blot analysis of the indicated proteins in circulating sEVs from melanoma mouse models. Circulating sEVs from equal volume of plasma (40 μL) were loaded per lane (n = 5). (F) Experimental schema of IBP and sEV treatment for the mice inoculated with B16F10 WT cells. (G) The growth curves of B16F10 tumors in C57BL/6 mice after five injections of sEVs from the indicated B16F10 with IBP treatment (n = 6). (H) The proportions of Ki67⁺CD45⁺CD8 or GzmB⁺CD45⁺CD8 TIL cells and Ki67⁺CD45⁺CD8 T cells in lymph nodes and spleens (n = 6). (I) Immunofluorescence staining of CD8 TILs in tumor tissues. Scale bar, 200 μm. (J) The number of CD8 TILs for each mouse, quantified from five high-power fields (HPFs) of tumor tissues (n = 6). Data are represented as mean ± SEM. Significance was determined using a two-tailed unpaired t test.

Figure 6

IBP potentiates anti-PD-1 therapy in OSCC and melanoma mouse models (A) Schematic illustration of the combined immunotherapy with IBP treatment in the B16F10 tumor-bearing mice. (B) The growth curves of B16F10 tumors in C57BL/6 mice subjected to the indicated treatments (n = 6). (C) The survival curves of the B16F10 tumor-bearing mice subjected to the indicated treatments (n = 6). (D) Schematic illustration of the establishment of the OSCC mouse model and therapeutic schedule. (E) Representative photographs of the tongues from the mice subjected to the indicated treatments. (F) The quantification of lesion areas (red areas) in the dorsal tongues. (G) The survival curves of the mice bearing OSCC subjected to the indicated treatments (n = 7). Data are represented as mean ± SEM. In (B) and (F), significance was determined using a two-tailed unpaired t test. In (C) and (G), significance was determined using the Mantel-Cox test.

Figure 7

Dermal topical administration of IBP enhances anti-PD-1 treatment in the melanoma mouse model (A) Representative photograph of the IBP-gel. (B and C) Analysis of the IBP release from the IBP-gel in PBS by HPLC. (D) Therapeutic schedule for the C57BL/6 mice inoculated with B16F10 WT or HRS KO cells. (E) Representative photograph of dermal topical administration using IBP-gel in the mice. The red dashed line delineates the tumor region. (F and G) The tumor growth (F) and survival curves (G) of B16F10 WT tumor-bearing mice subjected to the indicated treatments (n = 7). (H and I) The tumor growth (H) and survival curves (I) of the B16F10 HRS KO tumor-bearing mice subjected to the indicated treatments (n = 5). Data are represented as mean ± SEM. In (F) and (H), significance was determined using a two-tailed unpaired t test. In (G) and (I), significance was determined using the Mantel-Cox test.