ICAM-1-mediated Adhesion is a Prerequisite for Exosome-induced T Cell Suppression

SUMMARY

Tumor-derived extracellular vesicles (TEVs) suppress the proliferation and cytotoxicity of CD8⁺ T cells, thereby contributing to tumor immune evasion. Here we report that the adhesion molecule ICAM-1 co-localizes with PD-L1 on the exosomes; both ICAM-1 and PD-L1 are upregulated by interferon-γ. Exosomal ICAM-1 interacts with LFA-1, which is upregulated on activated T cells. Blocking ICAM-1 on TEVs reduces the interaction of TEVs with CD8⁺ T cells, and attenuates PD-L1-mediated suppressive effects of TEVs. During this study, we have established an Extracellular vesicle-Target cell Interaction Detection through SorTagging (ETIDS) system to assess the interaction between a TEV ligand and its target cell receptor. Using this system, we demonstrate that the interaction of TEV PD-L1 with PD-1 on T cells is significantly reduced in the absence of ICAM-1. Our study demonstrates that ICAM-1-LFA-1 mediated adhesion between TEVs and T cells is a prerequisite for exosomal PD-L1 mediated immune suppression.

Graphical Abstract

In brief

Tumor-derived exosomes can suppress the proliferation and cytotoxicity of CD8+ T cells. Zhang and colleagues report that the exosomes adhere to activated CD8 T cells through an ICAM-1-LFA-1 interaction, which is a prerequisite for exosomal PD-L1-PD-1 binding and T cell suppression.

INTRODUCTION

Immune checkpoint blockade (ICB)-based therapies, especially anti-PD-1 and anti-PD-L1 antibodies, have demonstrated impressive efficacy in the treatment of a variety of cancers (Baumeister et al., 2016; Chen and Mellman, 2017; Ribas and Wolchok, 2018; Wei et al., 2018). However, only a small portion of patients respond to these therapies (Hegde and Chen, 2020; Sharma et al., 2017). Further research is needed to better understand the molecular mechanisms of immune checkpoint-mediated immune suppression. Recent studies have demonstrated that tumor-derived extracellular vesicles (TEVs), especially exosomes, carry PD-L1 on their surfaces, and suppress the proliferation and activation of CD8⁺ T cells (Chen et al., 2018; Fan et al., 2019; Kim et al., 2019; Monypenny et al., 2018; Poggio et al., 2019; Ricklefs et al., 2018; Theodoraki et al., 2018; Yang et al., 2018). To better understand the molecular mechanisms by which TEVs exert their immune suppressive effect, it is important to elucidate the molecular basis of TEV-T cell interaction and how this interaction is regulated.

The interactions of T cells with endothelial cells, antigen-presenting cells (APCs) and cancer cells are important for their migration, maturation and function (Harjunpaa et al., 2019). Lymphocyte function-associated antigen-1 (LFA-1 or α_Lβ₂, CD11a/CD18) is a key integrin expressed on the surface of T cells, and ICAM-1 is a major ligand for LFA-1 (Walling and Kim, 2018). During immunological synapse formation, T cells interact with APCs through binding of LFA-1 to ICAM-1 at the center of the synapse prior to the interaction of T cell receptors (TCRs) with peptide-loaded major histocompatibility complex (pMHC) (Dustin, 2002). Furthermore, LFA-1 and its cognate ligands expressed on tumor cells play a critical role in T cell killing of tumor cells (Franciszkiewicz et al., 2013).

In addition to its expression on the cell surface, ICAM-1 has also been found on exosomes derived from dendritic cells (DC) and mast cells (Morelli et al., 2004; Segura et al., 2005; Skokos et al., 2001). ICAM-1 regulates the internalization of exosomes by immature DCs for antigen presentation to CD4⁺ T cells (Morelli et al., 2004). ICAM-1 on DC-derived exosomes is also required to prime naïve T cells (Segura et al., 2005). The high-affinity state of LFA-1 on activated CD4⁺ T cells is important for exosome binding (Nolte-'t Hoen et al., 2009). ICAM-1 was also found on tumor cell-derived exosomes, which interfere with the adhesion between leukocytes and activated endothelial cells by binding to the leukocytes (Lee et al., 2010). However, whether ICAM-1 on tumor cell-derived exosomes contributes to immunosuppression remains unknown.

In this study, we investigated the role of ICAM-1 in TEV-mediated suppression of CD8⁺ T cells in vitro and in vivo. We have also established an in vitro system, which allowed us to interrogate the interaction between exosomes and their target cells. With this system, the interaction of exosomal PD-L1 with T cell PD-1 can be analyzed. Our study indicates that ICAM-1 is essential for the interaction between exosomes and T cells, and that this interaction is a prerequisite for exosomal PD-L1-mediated immune suppression.

RESULTS

ICAM-1 is expressed on tumor-derived exosomes and its level is up-regulated by IFN-γ

EVs derived from human melanoma cell lines WM9 and WM164 were purified by differential centrifugation and verified by nanoparticle tracking analysis (NTA) (Figure 1A). By loading the same amounts of whole cell lysate, exosomes and microvesicles (MVs), we observed that ICAM-1 was enriched in exosomes rather than in MVs (Figure 1B). ICAM-1 co-fractionated with PD-L1 and other known exosome marker proteins including Alix, CD63 and Tsg101 as analyzed by iodixanol density gradient centrifugation (Figure 1C). Interferon-γ (IFN-γ) secreted by activated T cells upregulates tumor cell surface ICAM-1 (Hamai et al., 2008; Kim et al., 1995). We found that ICAM-1 in EVs, including exosomes and MVs, was also upregulated after IFN-γ treatment (Figure 1D and 1E). The degrees of ICAM-1 or PD-L1 upregulation were different for WM164 and WM9 cells, probably due to their different basal expression levels and mutational backgrounds. We also observed decreased levels of CD63 on exosomes derived from WM9 cells treated with IFN-γ. We next established an ELISA assay to assess the expression of ICAM-1 on TEVs (Figure 1F). Our analysis showed that IFN-γ treatment significantly increased ICAM-1 expression on WM9 (Figure 1G) and WM164 cell-derived exosomes and MVs (Figure 1H). The expression level of ICAM-1 was ~8 ng/μg in WM164 cell-derived exosomes and ~80 ng/μg EVs in WM9 cell-derived exosomes. These levels were higher than the expression levels of exosomal PD-L1 from the same cells (~0.15 ng/μg EVs) (Chen et al., 2018).

Figure 1. ICAM-1 is expressed on melanoma cell-derived exosomes and is up-regulated by IFN-γ.

(A) Characterization of exosomes purified from melanoma cells (WM9 and WM164) using nanoparticle tracking analysis (NTA). The X-axis represents diameters, and the Y-axis represents the concentration (particles/ml) of exosomes. (B) Western blot analysis of ICAM-1, PD-L1 and exosome markers (Hrs, Alix, Tsg101 and CD63) in the whole cell lysate (WCL), purified exosomes (EXO) and microvesicles (MV) from WM9 and WM164 cells. All lanes were loaded with equal amounts of total protein. (C) ICAM-1 co-fractionated with PD-L1 and exosome markers (Alix, CD63 and Tsg101) on density gradient centrifugation. (D) Immunoblot analysis of ICAM-1 and PD-L1 in control and IFN-γ-treated cells and their exosomes. All lanes were loaded with equal amounts of total protein. (E) Quantification of EV ICAM-1 expression in WM9 (left panel) and WM164 (right panel) cells with or without IFN-γ treatment. (F) Schematic of ELISA to measure ICAM-1 levels on the surface of exosomes isolated from human melanoma cell supernatants. TMB, 3, 3′, 5, 5′-tetramethylbenzidine; SA-HRP, streptavidin-horseradish peroxidase. See MATERIALS AND METHODS for details. ELISA of ICAM-1 on exosomes and microvesicles from WM9 (G) and WM164 (H) melanoma cells, with or without IFN-γ treatment. Exosomes and MVs from WM9 and WM164 cells were collected and added into the ELISA plate which was coated with anti-ICAM-1 antibodies. Data represent mean ± s.d. of three independent biological replicates. Statistical analysis is performed using two-sided unpaired t-test (E, G, H).

Hrs and Rab27 mediate the secretion of exosomal ICAM-1

Exosomes are generated through a defined intracellular trafficking pathway that involves the endosomal sorting complexes required for transport (ESCRT) (Juan and Furthauer, 2018). As a pivotal member of ESCRT-0, Hrs mediates the initial recognition and sorting of protein cargo to the interluminal vesicles in the multivesicular bodies, which ultimately fuse with the plasma membrane for exosome release (Schmidt and Teis, 2012). Hrs knockdown led to a decrease of ICAM-1 level in exosomes, and accumulation of ICAM-1 in WM9 cells (Figure 2A). Immunofluorescence showed a co-localization of ICAM-1 with Hrs in the cell (Figure 2B). ICAM-1 also co-immunoprecipitated with Hrs from cell lysates (Figure 2C). In addition to Hrs, the small GTPase, Rab27A, mediates exosome release from cells (Ostrowski et al., 2010). Knockdown of Rab27A also blocked exosomal ICAM-1 secretion (Figure 2D). Besides western blotting, NTA data also demonstrated that knockdown of Hrs or Rab27A decreased the exosomes release from the cells (Figure S1). Together, these results further confirmed the exosomal trafficking of ICAM-1.

Figure 2. Hrs and Rab27 mediate the secretion of exosomal ICAM-1.

(A) Immunoblot analysis of ICAM-1, PD-L1 and exosome markers (CD63 and Tsg101) in Hrs-knockdown cells (left panel). Quantification of exosomal ICAM-1 is shown in the right panel. (B) Immunofluorescence staining showing the intracellular co-localization of ICAM-1 with Hrs in WM9 cells. Scale bars, 10 μm. (C) Co-immunoprecipitation of ICAM-1 with Flag-tagged Hrs from 293T cells expressing ICAM-1 and Hrs. (D) Immunoblot analysis of ICAM-1, PD-L1 and exosome markers (CD63 and Tsg101) in Rab27A-knockdown cells (left panel). Quantification of exosomal ICAM-1 is shown in the right panel. Data represent mean ± s.d. of three independent biological replicates. Statistical analysis is performed using one-way ANOVA analysis with Dunnett’s multiple comparison tests (A, B).

ICAM-1 and PD-L1 co-localize on TEVs

Next, we tested whether ICAM-1 and PD-L1 co-localized on the same exosomes. Using immuno-electron microscopy (EM), we observed ICAM-1 and PD-L1 on the same exosomes derived from WM9 cells (Figure 3A). We often saw more ICAM-1 proteins than PD-L1 on the exosomes, consistent with the above ELISA results. Using a single vesicle capture platform (ExoView), we detected PD-L1 ICAM-1 double positive exosomes. IFN-γ treatment significantly increased the proportion of the PD-L1⁺ ICAM-1⁺ exosomes among the total PD-L1⁺ exosomes (Figure 3B and 3C). In addition, using Dynabeads conjugated with anti-ICAM-1 antibodies, we precipitated ICAM-1⁺ exosomes from the small vesicles derived from cells (Figure 3D). Removal of ICAM-1⁺ exosomes markedly reduced the PD-L1 level on the remaining exosomes, and the captured ICAM-1⁺ exosomes showed high levels of PD-L1 (Figure 3E and 3F), suggesting that ICAM-1 and PD-L1 double positive exosomes are well presented in melanoma cell-derived exosomes. Finally, we analyzed the ICAM-1 level on PD-L1⁺ exosomes from the plasma of healthy donors and patients with metastatic melanoma. PD-L1⁺ exosomes were captured with an anti-PD-L1 antibody, and ICAM-1 on these exosomes was analyzed by ELISA (Figure 3G). The levels of ICAM-1 on PD-L1⁺ exosomes were higher in melanoma patients (Figure 3H).

Figure 3. ICAM-1 and PD-L1 colocalize on the exosomes derived from melanoma cells.

(A) An EM image of a WM9 cell-derived exosome co-stained with anti-ICAM-1 antibodies (conjugated with 10 nm gold particles) and anti-PD-L1 antibodies (conjugated with 5 nm gold particles). Scale bar, 100 nm. (B) ExoView images of exosomes derived from control and IFN-γ-treated WM9 cells. The exosomes were captured by anti-PD-L1 antibodies and then incubated with fluorescence-labeled anti-PD-L1 (green), anti-ICAM-1 (red) and anti-pantetraspanin (Tetra, a mixture of anti-CD63, anti-CD81 and anti-CD9, blue) antibodies. (C) Quantification of PD-L1⁺ ICAM-1⁺ exosomes among total Tetra⁺ exosomes. (D) Schematic of ICAM-1⁺ exosome isolation from WM9-derived TEVs by magnetic beads. See MATERIALS AND METHODS for details. (E) Immunoblot analysis of ICAM-1, PD-L1 and other exosome markers in total, ICAM-1⁺ and void exosomes from WM9 cells. The void and ICAM-1⁺ exosomes were isolated from the WM9 cells-derived exosomes. The same amounts of proteins were loaded in each lane. See MATERIALS AND METHODS for details. (F) Quantification of exosomal ICAM-1 (left) and PD-L1 (right) expression in WM9-derived exosomes sorted as shown in (E). (G) Schematic of ELISA to measure ICAM-1 levels on the surface of PD-L1⁺ exosomes isolated from melanoma patient plasma. (H) ELISA of ICAM-1 expression levels on circulating PD-L1⁺ exosomes in healthy donors (“HD”, n = 10) and melanoma patients (“MP”, n =27). Data represent mean ± s.d. of three independent biological replicates. Statistical analysis is performed using two-sided unpaired t-test (C), one-way ANOVA analysis with Dunnett’s multiple comparison tests (F) and two-sided unpaired Welch’s t-test (H).

ICAM-1 mediates the interaction of TEVs to CD8⁺ T cells through binding to LFA-1

Exosomes collected from control or IFN-γ treated WM9 cells were labeled with carboxyfluorescein succinimidyl ester (CFSE) and incubated with CD8⁺ T cells. CFSE-labeled exosomes bound to T cells stimulated by anti-CD3/CD28 antibodies, whereas minimal binding was detected with the unstimulated CD8⁺ T cells (Figure 4A). In addition, exosomes showed stronger T cell binding than MVs (Figure 4B). Pre-treatment of exosomes with anti-ICAM-1 antibodies markedly decreased T cell-exosome binding (Figure 4C). Knockdown of ICAM-1 in WM9 cells led to decreased expression of ICAM-1 on the exosomes without affecting PD-L1 expression (Figure S2A and S2C). Nanoparticle tracking analyses of WM9 cell culture supernatant after removal of dead cells and cell debris showed that ICAM-1 knockdown did not affect the size distribution of either MVs or exosomes (Figure S2D). Bradford assays confirmed that ICAM-1 knockdown did not affect the total levels of proteins on the exosomes (Figure S2E and S2F). Similar to the results from the anti-ICAM-1 antibody blocking assay, knockdown of ICAM-1 in WM9-derived exosomes showed reduced binding to the stimulated CD8⁺ T cells (Figure 4D).

Figure 4. ICAM-1/LFA-1 interaction mediates the adhesion between TEVs and stimulated CD8⁺ T lymphocytes.

(A) Representative histograms of human peripheral CD8⁺ T cells binding to CFSE-exosomes derived from WM9 cells with or without IFN-γ treatment. The proportions of CFSE positive CD8⁺ T cells (with or without stimulation with anti-CD3/CD28 antibodies) are shown on the right. (B) Representative histograms of human peripheral CD8⁺ T cells bound to CFSE-exosomes or microvesicles derived from WM9 cells with or without IFN-γ treatment. CD8⁺ T cells were stimulated with anti-CD3/CD28 antibodies. The proportions of CFSE positive CD8⁺ T cells are shown on the right. (C) Representative histograms of human peripheral CD8⁺ T cells bound to CFSE-exosomes that were pre-treated with IgG or anti-ICAM-1 antibodies. The proportions of CFSE positive CD8⁺ T cells are shown on the right. (D) Representative histograms of human peripheral CD8⁺ T cells treated with CFSE exosomes derived from control or ICAM-1 knockdown (KD) WM9 cells. The proportions of CFSE positive cells are shown on the right. (E) Representative histograms of human peripheral CD8⁺ T cells that bound to CFSE-exosomes pre-treated with anti-CD11a, anti-CD11b or anti-CD18 antibodies, then treated with exosomes from WM9 cells. The proportions of CFSE positive CD8⁺ T cells are shown on the right. Data represent mean ± s.d. of three independent biological replicates. Statistical analysis is performed using unpaired t-test (A) or one-way ANOVA analysis with Dunnett’s (B, D) or Sidak’s (C, E) multiple comparison tests.

Two major receptors of ICAM-1 have been found in immune cells, LFA-1 (CD11a/CD18) and Macrophage-1 antigen (Mac-1; CD11b/CD18) (Schmidt and Teis, 2012). To determine the receptor that interacts with exosomal ICAM-1, antibody blocking experiments were performed. Pre-treatment of CD8 T cells with anti-CD11a or anti-CD18 antibodies, but not anti-CD11b antibodies, reduced the binding of TEVs to CD8⁺ T cells (Figure 4E). Flow cytometry analyses showed that CD11a and CD18, but not CD11b, were significantly upregulated on the surface of stimulated CD8⁺ T cells (Figure S3A and S3B). Using an antibody that specifically recognizes the activated form of LFA-1, we confirmed that activated LFA-1 was significantly increased in stimulated CD8⁺ T cells (Figure S3A and S3B). These results suggest that exosomal ICAM-1 mainly interacts with activated LFA-1 on stimulated CD8⁺ T cells, which also explains why exosomes have stronger binding to stimulated CD8⁺ T cells compared to unstimulated CD8⁺ T cells.

ICAM-1 is needed for the efficient inhibition of CD8⁺ T cells by TEVs

We have previously shown that TEVs exert an inhibitory effect on stimulated CD8⁺ T cells (Li et al., 2019). We thus examined the role of ICAM-1 in this inhibition. Genetically engineered Jurkat T cells with NFAT-mediated activation of luciferase activities were used as effector cells and generated luminescence upon stimulation by anti-CD3/CD28 antibodies (Figure 5A). Luminescence was significantly reduced by WM9-derived exosomes (Figure 5B). TEVs pre-treated with anti-ICAM-1 antibodies or effector cells pre-blocked with anti-CD18 antibodies rescued the activation of the effector cells (Figure 5B), suggesting that the interaction of ICAM-1 on TEVs with LFA-1 mediates T cell inhibition. Using CFSE labeling (cell division-tracking dye) and granzyme B (GzmB) expression, we previously showed that exosomes from WM9 cells inhibited the proliferation and cytotoxicity of stimulated CD8⁺ T cells (Chen et al., 2018). Using the same assay, we found that pre-treating TEVs with anti-ICAM-1 antibodies significantly attenuated their inhibitory effects (Figure 5C and 5D). Similarly, TEVs with ICAM-1 knockdown (ICAM-1 KD) showed much weaker T cell inhibition than control cells, as indicated by the higher levels of Ki67 and GzmB expression in the stimulated CD8⁺ T cells treated with ICAM-1 KD exosomes (Figure 5E and 5F). WM9-derived exosomes also inhibited the secretion of cytokines, including IL-2, TNF-α and IFN-γ from stimulated CD8⁺ T cells. Pre-treating exosomes with anti-ICAM-1 antibodies, but not IgG isotype, significantly attenuated the inhibitory effects (Figure S4A-S4C).

Figure 5. ICAM-1 is required for T cell suppression by TEVs.

(A) Schematic of the exosomal ligand-receptor blockade assay. Genetically engineered Jurkat T cells with NFAT-mediated expression of luciferase activities were stimulated with anti-CD3/CD28 antibodies. TEVs derived from WM9 cells were pre-treated with anti-ICAM-1 antibodies and added into the system. Luminescence was measured to indicate the activation of effector T cells. (B) Effector T cell activation with indicated treatment. RLU, relative luminometer units. (C) Histogram of CFSE positive human peripheral CD8⁺ T cells treated with WM9-derived exosomes. The exosomes were pre-treated with IgG isotype or anti-ICAM-1 antibodies. The proportions of cells with diluted CFSE dye (proliferating cells) are shown on the right. (D) Representative contour plots of human peripheral CD8⁺ T cells with indicated treatment for the expression of granzyme B (GzmB). The proportions of GzmB positive CD8⁺ T cells are shown on the right. (E) Representative contour plots of human peripheral CD8⁺ T cells for the expression of Ki67 after indicated treatment. The proportions of Ki67 positive CD8⁺ T cells are shown on the right. (F) Representative contour plots of human peripheral CD8⁺ T cells for the expression of GzmB after indicated treatment. The proportions of GzmB positive CD8⁺ T cells are shown on the right. Data represent mean ± s.d. of three independent biological replicates. Statistical analysis is performed using one-way ANOVA with Sidak’s multiple comparison tests (B, C, D, E, F).

In addition to melanoma, we also measured the exosomal ICAM-1 derived from other cancer cells. EVs derived from colon cancer cell HCT116 contained ICAM-1 and PD-L1 (Figure S5A and S5B). Their expression in exosomes was up-regulated by IFN-γ (Figure S5C and S5D). Pre-treatment of exosomes with anti-ICAM-1 antibodies significantly decreased the T cell-exosome binding (Figure S5E). HCT116-derived exosomes inhibited Ki67 and GzmB expression in CD8⁺ T cells, and pre-treatment with anti-ICAM-1 antibodies blocked the effects (Figure S5F and S5G). The same observations were made for EVs derived from human lung cancer cell lines, H1264 and H1299 (Figure S6).

PD-L1 is not required for the adhesion of TEVs to stimulated CD8⁺ T cells

PD-L1 has been shown to be important for TEV-mediated inhibition of CD8⁺ T cells (Chen et al., 2018; Ricklefs et al., 2018). We tested whether exosomal PD-L1 was necessary for TEV-T cell adhesion. WM9 cells with PD-L1 knockout were generated using a CRISPR/Cas9 system. There was no PD-L1 expression in both the whole cell lysate and exosomes derived from the KO cells, while the expression of ICAM-1 was not affected (Figure 6A). The PD-L1 KO TEVs were collected and labeled with CFSE to perform the adhesion assays. PD-L1 depletion did not affect the adhesion between TEVs and CD8⁺ T cells (Figure 6B). On the other hand, PD-L1 KO TEVs showed weakened suppression of CD8⁺ T cells, as demonstrated by the CFSE proliferation assay, Ki67 and GzmB expression (Figure 6C-6E). These results suggest that, while TEV PD-L1 functionally inhibits CD8⁺ T cells, its effect on the adhesion between TEVs and T cells is minimal.

Figure 6. Exosomal PD-L1 is not necessary for the binding of TEVs to CD8⁺ lymphocytes.

(A) Immunoblot analysis of PD-L1, ICAM-1 and exosome markers (CD63, Hrs and CD9) in the whole cell lysate (WCL) and purified exosomes (EXO) from control and PD-L1 knockout (KO) WM9 cells. The same amounts of proteins were loaded in each lane. (B) Representative histograms of human peripheral CD8⁺ T cells bound to CFSE-exosomes from control and PD-L1 KO WM9 cells. The proportions of exosome-binding CD8⁺ T cells are shown on the right. (C) Representative histogram of CFSE-labelled human peripheral CD8⁺ T cells after indicated treatment. The proportions of cells with diluted CFSE dye (proliferating cells) are shown on the right. (D) Representative contour plots of human peripheral CD8⁺ T cells for the expression of Ki67 after indicated treatment. The proportions of Ki67 positive CD8⁺ T cells are shown on the right. (E) Representative contour plots of human peripheral CD8⁺ T cells for the expression of granzyme B (GzmB) after indicated treatment. The proportions of GzmB positive CD8⁺ T cells are shown on the right. Data represent mean ± s.d. of three independent biological replicates. Statistical analysis is performed using two-sided unpaired t-test (B) or one-way ANOVA analysis with Sidak’s multiple comparison tests (C, D, E).

ICAM-1 is necessary for TEV-induced immunosuppression in vivo

We then tested the role of exosomal ICAM-1 in CD8⁺ T cell suppression in vivo. We used YUMM1.7 mouse melanoma cell in our study, as ICAM-1 was reported to be expressed on its surface (Nakayama et al., 1997). We knocked out ICAM-1 in YUMM1.7 cells (YUMM1.7 ICAM-1 KO) using the CRISPR/Cas9 system, which ablated ICAM-1 from the exosomes without affecting PD-L1 expression (Figure 7A). These exosomes had weaker binding to the stimulated mouse splenic CD8⁺ T cells (Figure 7B). Blocking ICAM-1 with anti-ICAM-1 antibodies reduced the binding of control YUMM1.7-derived exosomes to the mouse splenic CD8⁺ T cells (Figure 7C). Exosomes derived from YUMM1.7 cells also inhibited the proliferation and cytotoxicity of mouse splenic CD8⁺ T cells. Pre-treatment of the exosomes with anti-ICAM-1 antibodies or treatment of mouse splenic CD8⁺ T cells with ICAM-1 KO exosomes reduced these effects (Figure 7D).

Figure 7. Blocking exosomal ICAM-1 inhibits T cells and promotes tumor progression in vivo.

(A) Immunoblot analysis of PD-L1, ICAM-1 and exosome-associated markers (Tsg101 and CD9) in the whole cell lysate (WCL) and purified exosomes (EXO) from control and ICAM-1 knockout (KO) YUMM1.7 mouse cell line. The same amounts of proteins was loaded in each lane. (B) Representative histograms of splenic CD8⁺ T cells (stimulated with anti-CD3/CD28 antibodies) bound to CFSE-labeled exosomes derived from control and ICAM-1 KO YUMM1.7 cells. The proportions of exosome-bound CD8⁺ T cells are shown on the right. (C) Representative histograms of splenic CD8⁺ T cells bound to YUMM1.7-derived exosomes pre-treated with IgG isotype or anti-ICAM-1 antibodies. The proportions of exosome-bound CD8⁺ T cells are shown on the right. (D) Representative contour plots of Ki-67 and granzyme B (GzmB) expression in stimulated mouse splenic CD8⁺ T cells after treatment with exosomes from control or ICAM-1 KO YUMM1.7 cells, or from YUMM1.7 cell-derived exosomes with or without pre-treatment of IgG isotype or anti-ICAM-1 antibodies. The percentage of Ki67⁺GzmB⁺ CD8⁺ T cells among total is shown on the right. (E) Growth curve of YUMM1.7 tumors in mice with tail vein injections with indicated exosomes (n = 7 mice per group). (F) The number of tumor infiltrating CD8⁺ lymphocytes (TILs) from the YUMM1.7 tumors with indicated treatment were analyzed using flow cytometry 22 days post-implantation. The expression levels of Ki67 (G) and GzmB (H) by PD-1⁺ TILs as in (E) were analyzed using flow cytometry. Data represent mean ± s.d. of three independent biological replicates. Statistical analysis is performed using two-sided unpaired t-test (B), one-way ANOVA analysis with Sidak’s multiple comparison tests (D), two-way ANOVA analysis with Tukey’s multiple comparison tests (E), or Welch ANOVA with Dunnett’s T3 multiple comparison tests (F, G, H).

Next, we established a syngeneic C57BL/6 mouse model using YUMM1.7 cells. Infusion of purified exosomes derived from YUMM1.7 cells promoted the growth of tumors, whereas exosomes from ICAM-1 KO YUMM1.7 cells or the exosomes pre-treated with anti-ICAM-1 antibodies attenuated this effect (Figure 7E). The tumors were harvested 22 days post-implantation. The infiltration and activation of CD8⁺ T cells in tumor tissues were analyzed using flow cytometry (Figure S7). The number of tumor-infiltrating CD8⁺ T lymphocytes (TILs) decreased after the injection of YUMM1.7 exosomes or exosomes pre-treated with the IgG isotype antibodies. However, this was not seen following the injection of exosomes from YUMM1.7 ICAM-1 KO cells or those pretreated with anti-ICAM-1 antibodies (Figure 7F). Depletion of ICAM-1 on exosomes or pretreating exosomes with anti-ICAM-1 antibodies decreased the inhibitory effects on the proliferation and cytotoxicity of infiltrating PD-1⁺ CD8⁺ T cells in tumor (Figure 7G and 7H).

In addition to the melanoma YUMM1.7 cells, we also examined the expression of exosomal ICAM-1 and PD-L1 derived from murine colon cancer cell line MC38. EVs, especially exosomes, from MC38 cells contained ICAM-1 and PD-L1 (Figure S8A and S8B), and their expression was upregulated by IFN-γ (Figure S8C and S8D). Pre-treatment with anti-ICAM-1 antibodies also inhibited the binding between murine splenic CD8⁺ T lymphocytes and MC38-derived exosomes (Figure S8E). MC38-derived exosomes decreased the Ki67⁺ GzmB⁺ population of CD8⁺ T cells, and pre-treatment with anti-ICAM-1 antibodies suppressed the effects (Figure S8F). We also established a syngeneic C57BL/6 mouse model using MC38 cells. Infusion of purified exosomes derived from MC38 cells promoted the growth of tumors, while exosomes pre-treated with anti-ICAM-1 antibodies attenuated this effect (Figure S8G). The infiltration and activation of CD8⁺ T cells in tumor microenvironment were analyzed 22 days post-implantation of tumors. Immunohistochemistry (IHC) staining showed CD8⁺ TILs was decreased after the injection of exosomes pre-treated with the IgG isotype antibodies, but not the exosomes pre-treated anti-ICAM-1 antibodies (Figure S8H and S8I). Flow cytometry analysis also showed the decreased CD8⁺ TILs after the injection of exosomes pre-treated with IgG isotype; pre-treating exosomes anti-ICAM-1 antibodies showed less inhibitory effect (Figure S8J). Finally, pretreating exosomes with anti-ICAM-1 antibodies also blocked the inhibitory effects of exosomes on the proliferation (as shown by Ki67 expression, Figure S8K) and cytotoxicity (as shown by Granzyme B expression, Figure S8L) of infiltrating PD-1⁺ CD8⁺ T cells.

The necessity of ICAM-1 in exosomal PD-L1 interaction with PD-1 on T cell surface as tested by the ETIDS system

Studying the interaction of an exosome protein with its ligand on the surface of a recipient cell is important but challenging. The Staphylococcus aureus transpeptidase sortase A (SrtA) has been utilized to probe the interaction of a ligand and its receptor in two cells based on proximity-dependent labelling across cell-cell interfaces (Pasqual et al., 2018). SrtA covalently transfers a substrate containing the sorting motif “LPXTG” to a tag fused to a receptor that consists of five N-terminal glycine residue (“G5”), enabling the detection of the interaction between a ligand and receptor across the cells (Pasqual et al., 2018). To examine the potential interaction between EVs and their target cells, we established an Extracellular vesicle-T cell Interaction Detection through SorTagging (ETIDS) system (Figure 8A). We expressed murine PD-L1 that fused to SrtA in WM9 cells and murine PD-1 with a G5 tag in Jurkat T cells. Exosomes from WM9 cells with mPD-L1-SrtA were purified and confirmed by western blotting (Figure 8B). When mPD-L1-SrtA WM9 cells and their exosomes were incubated with biotinylated substrate (biotin-LPETG), mPD-L1-SrtA formed an acyl intermediate due to conjugation with biotin-LPETG (Figure 8C). We next incubated G5-mPD-1 Jurkat T cells with exosomes from mPD-L1-SrtA WM9 cells in the presence of biotin-LPETG. Flow cytometry analysis showed that G5-PD-1 Jurkat T cells were biotinylated only when incubated with the mPD-L1-SrtA⁺ exosomes in the presence of biotin-LPETG (Figure 8D). To block the PD-L1-PD-1 interaction, anti-human or anti-mouse PD-L1 antibodies were added into the co-culture system and the biotinylated Jurkat T cells were analyzed. Anti-mouse PD-L1 antibodies markedly reduced the biotinylated population of the Jurkat T cells (Figure 8E). The anti-human PD-L1 antibodies did not affect the biotinylation, consistent with the finding that the PD-L1-PD-1 interaction was not necessary for exosome-T cell association shown in Figure 6. Pre-treatment of PD-L1-SrtA⁺ exosomes with anti-human ICAM-1 antibodies, but not anti-human CD63 antibodies, significantly reduced the biotinylated population of Jurkat cells (Figure 8F). We then knocked down ICAM-1 expression in mPD-L1-SrtA WM9 cells and collected the exosomes to measure the interaction of exosomal PD-L1 and Jurkat PD-1. ICAM-1 knockdown significantly reduced the biotin-positive Jurkat cells (Figure 8G). Together, these results suggest that ICAM-1 is needed for interaction between exosome PD-L1 and PD-1 on T cell surface.

Figure 8. Establishing the ETIDS system and testing the necessity of ICAM-1 in exosomal PD-L1 interaction with PD-1 on T cell surface.

(A) Schematic representation of the ETIDS system. Murine PD-L1-SrtA and murine PD-1-G5 were stably expressed in WM9 and Jurkat T cells, respectively. PD-L1-SrtA⁺ exosomes were collected from WM9 cells and incubated with Jurkat PD-1-G5 cells in the presence of Biotin-LPETG. Biotin-LPETG are transferred to G5 tag on Jurkat cells that are catalyzed by SrtA on exosomes. The biotin signals were analyzed using flow cytometry. See MATERIALS AND METHODS for details. (B) Western blot analysis showing the expression of PD-L1-SrtA in whole cell lysates and exosomes derived from WM9 PD-L1-SrtA cells. (C) Western blot analysis showing PD-L1-SrtA on exosomes was conjugated with biotin-LPETG. (D) Flow cytometry analysis of Jurkat T cells after treatment with WM9 PD-L1-SrtA-derived exosomes. (E) Flow cytometry analysis of biotin positive G5-PD-1 Jurkat T cells after incubation with PD-L1-SrtA⁺ exosomes that were pre-treated with anti-human PD-L1 or anti-mouse PD-L1 antibodies. The proportions of biotin positive Jurkat T cells are shown on the right. (F) Flow cytometry analysis of biotin positive G5-PD-1 Jurkat T cells after incubation with PD-L1-SrtA⁺ exosomes that were pre-treated with IgG isotype, anti-human CD63 antibodies, or anti-human ICAM-1 antibodies. The proportions of biotin positive Jurkat T cells are shown on the right. (G) Flow cytometry analysis of biotin positive G5-PD-1 Jurkat T cells after treatment with PD-L1-SrtA⁺ exosomes that were harvested from ICAM-1 KD or control WM9 cells. The proportions of biotin positive Jurkat T cells are shown on the right. Data represent mean ± s.d. of three independent biological replicates. Statistical analysis is performed using one-way ANOVA analysis with Sidak’s (E, F) or Dunnett’s multiple comparison tests (G).

DISCUSSION

Molecular study of tumor-derived EVs is fundamental to our understanding of their ability to suppress the immune system. Our study established a role of ICAM-1 in the immunosuppressive function of PD-L1⁺ EVs. 1) ICAM-1 was expressed on tumor EVs, mostly exosomes; immune EM imaging showed that exosomal ICAM-1 had the same membrane topology as cell surface ICAM-1, with the extracellular adhesion domain exposed on the surface; 2) ICAM-1 and PD-L1 were co-expressed on the exosomes, and both proteins were up-regulated by IFN-γ; 3) ICAM-1 antibody blocking or RNAi knockdown reduced the interaction between TEVs and CD8⁺ T cells, and attenuated the suppressive effect of TEVs on CD8⁺ T cell proliferation and activation in vitro. In contrast, antibody blocking or CRISPR KO of PD-L1, while attenuated the inhibition on CD8⁺ T cells, did not affect the binding of TEVs to CD8⁺ T cells; 4) Blocking ICAM-1 diminished the tumor-promoting effect of EVs and increased the infiltration and activation of PD-1⁺ CD8⁺ T cells in the tumor.

EVs affect the pathophysiology of their recipient cells through two different modes (Kalluri and LeBleu, 2020; Maia et al., 2018; McKelvey et al., 2015; Morrissey and Yan, 2020; Tkach and Thery, 2016; Turturici et al., 2014). The first involves the internalization of EVs by the recipient cells for the horizontal transfer of bioactive materials (e.g. signaling proteins and mRNAs), and the second is mediated by ligand-receptor interaction at the cell surface (Colombo et al., 2014; Kalluri and LeBleu, 2020; Tkach and Thery, 2016). Adhesion molecules such as integrins are essential for both types of interactions (Colombo et al., 2014; Hoshino et al., 2015; Tkach and Thery, 2016). Previous studies have identified ICAM-1 on exosomes derived from mast cells and dendritic cells, and demonstrated a role of ICAM-1 in naïve T cell priming and activation (Morelli et al., 2004; Nolte-'t Hoen et al., 2009). These studies implicated exosome engulfment and subsequent transfer of molecules to the recipient lymphocytes in their activation. For the effect of tumor-derived exosomes on stimulated cytolytic T cells, while signaling molecule transfer via exosome internalization cannot be totally ruled out, the observed suppressive effect was mostly mediated by inhibitory ligand-receptor interactions (e.g. the immune checkpoint signaling) on the cell surface. Whiteside and colleagues reported that exosomes tended to adhere to the surface of CD8⁺ T cells rather than being internalized (Muller et al., 2017). We and others found that EVs derived from metastatic cancer cells inhibited the proliferation and activation of CD8⁺ T cells through the interaction of PD-L1 on TEVs with PD-1 on CD8⁺ T cells (Chen et al., 2018; Poggio et al., 2019; Yang et al., 2018). Here, we propose that ICAM-1-mediated adhesion of TEVs to CD8⁺ T cells is a prerequisite for the PD-L1/PD-1-mediated suppression of CD8⁺ T cells by TEVs. PD-L1 and ICAM-1 are co-expressed on EVs and co-upregulated by IFN-γ, which allows exosomal ICAM-1 to act in a coordinated fashion with PD-L1 in CD8⁺ T cell suppression. ICAM-1 binds to LFA-1 with a high affinity and slow dissociation rate, providing a physical force for adhesion between cells (Tominaga et al., 1998). On the other hand, the interaction between PD-L1 and PD-1 is relatively weak (Li et al., 2017). Our ELISA data show that the expression levels of ICAM-1 were much higher on TEVs than that of PD-L1, further suggesting that the binding of CD8⁺ T cells to TEVs is more likely to be mediated by the ICAM-1-LFA-1 interaction. Supporting this model, blocking the ICAM-1-LFA-1 interaction reduced the binding of TEVs to CD8⁺ T cells and attenuated the inhibitory effect of TEVs, whereas blocking the interaction between PD-L1 and PD-1 attenuated the inhibitory function of TEVs to CD8⁺ T cells without affecting the adhesion of CD8⁺ T cells to TEVs. Besides ICAM-1, other candidate adhesion proteins such as CEACAM1 and ICAM-4, which are upregulated in cancer cells in response to IFN-γ (Markel et al., 2009; Rusinova et al., 2013), may also contribute to the binding between activated CD8⁺ T cells and TEVs.

It is interesting to note that the level of LFA-1 is upregulated on activated CD8⁺ T cells (Castro et al., 2018; Kasahara et al., 1983), which secrete higher levels of IFN-γ. As IFN-γ in turn upregulates the expression of ICAM-1 on the exosomes, the reciprocal upregulation of the cognate pair of adhesion molecules allows the exosomes to better interact with stimulated CD8⁺ T cells. This may offer an explanation why exosomes generated from tumor cells efficiently recognize and inhibit the activated CD8⁺ T cells.

During this study, we have established the ETIDS platform to assay the interaction of exosomes with T cells. This system allows us to detect the specific interaction between exosomal PD-L1 and T cell PD-1, and interrogate the role of ICAM-1. SrtA uses a peptide substrate that is easily synthesized and can be linked to a variety of detectable labels, which simplifies detection. Most importantly, as the detection is based on the transfer of a short peptide labeled with biotin, it allows us to “record” both the on-going and past interactions. We believe that this assay could be used to the study of a wide range of EV-target cell interactions in the field.

Our study suggests that TEVs act as a precisely equipped functional unit for their interaction with, and checkpoint inhibition of, CD8⁺ T cells. Interestingly, both CD11a (a subunit LFA-1) and PD-1 have been used to identify circulating tumor-reactive CD8⁺ T cells in peripheral blood of patients with metastatic diseases (Evans et al., 2019; Liu et al., 2013). The presence of circulating ICAM-1 and PD-L1 positive EVs reported here provide a well matched regulatory mechanism that reshapes antitumor immunity. Further characterization of the molecular architecture of TEVs will be important for the mechanistic understanding of their function, and may guide the effort to develop EV-based therapeutical carriers that effectively modulate the immune system in cancer patients.

Limitations of Study

While we focus on ICAM-1 in this study, some other adhesion molecules, such as CEACAM1 and ICAM-4, could also mediate the interaction between TEVs and CD8⁺ T cells. We have used melanoma cells, lung cancer cells, colon cancer cells in our study. It would be interesting to expand the studies to other cancer types.

STAR METHODS

RESOURCE AVAILABILITY

○ Lead contact

Further information and requests for resources and reagents should be directed to the Lead Contact, Wei Guo (guowei@sas.upenn.edu).

Materials availability

Plasmid and cell lines generated for this work is available upon request.

Data and code availability

• Original western blot images and microscope data reported in this paper are available upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mammalian cell lines

The human melanoma cell lines WM9 and WM164 presented in this study were established in Meenhard Herlyn’s laboratory at The Wistar Institute. The murine melanoma cell line YUMM1.7 was originally obtained from Marcus Bosenberg (Yale University) (Meeth et al., 2016). The human lung cancer cell lines H1264 and H1299 were obtained from Dr Steven Albelda’s laboratory (University of Pennsylvania). The human colon cancer cell line HCT116 were generously provided by Dr Shiaw-Yih Lin’s laboratory (MD Anderson Cancer Center). The murine colon cancer cell line MC38 was purchased from ATCC. All cell lines were authenticated by DNA fingerprinting and tested routinely before use in order to avoid mycoplasma contamination. All cells were cultured at 37°C with humidified 5% CO₂. Human melanoma cells, lung cancer cells and colon cancer cells were cultured in RPMI1640 medium (Gibco, CA, USA) with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin. YUMM1.7 cells were maintained in DMEM/F12 media containing 10% FBS, 1% non-essential amino acids and 1% penicillin–streptomycin. MC38 cells were maintained in DMEM containing 10% FBS and 1% penicillin–streptomycin. For stimulation of IFN-γ, cells were incubated with 100 ng/ml of recombinant human or mouse IFN-γ (Peprotech) for 48 h.

Mice

6-8 weeks old female mice were used for all experiments. C57BL/6 wide type mice were purchased from The Jackson Laboratory. Prior to all experiments, purchased mice were allowed one week to acclimate to housing conditions at the University of Pennsylvania Perelman School of Medicine animal facility. All experimental mice were housed in specific pathogen–free conditions and all animal experiments were performed according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Pennsylvania.

Patient plasma

Patient peripheral blood samples were collected in K2 EDTA tubes from Stage III or IV melanoma patients. Plasma was isolated within 24 h of blood collection. All patients provided informed consent for blood collection under the University of Pennsylvania Abramson Cancer Center’s melanoma research program tissue collection protocol UPCC# 08607 / IRB# 703001 approved by the Institutional Review Board of the University of Pennsylvania. Blood samples from healthy donors were collected at The Wistar Institute after approval by the Institutional Review Board. The gender and age of the individuals are shown in Supplemental Table 1.

METHOD DETAILS

shRNA and CRISPR/Cas9 genome editing

The shRNAs against human HRS (also known as HGS) (NM_004712, GCACGTCTTTCCAGAATTCAA, GCATGAAGAGTAACCACAGC), human RAB27A (NM_004850.3, GCTGCCAATGGGACAAACATA, CAGGAGAGGTTTCGTAGCTTA) (gift from A. Weaver, Vanderbilt University), and the shRNAs against human ICAM-1 (also known as CD54) (NM_000201, GCCAACCAATGTGCTATTCAA, CCTCAGCACGTACCTCTATAA) (obtained from the Wistar institute) or scrambled shRNA (Addgene #1864) were packaged into lentiviral particles using 293T cells co-transfected with viral packaging plasmids. Lentiviral supernatants were harvested 48-72 h after transfection. WM9 cells were infected with filtered lentivirus and selected with 2 μg/ml puromycin. Knockdown was verified by western blotting.

The gRNA oligonucleotides against human PD-L1 (sgRNA 1, 5’-CCTTGCACTTCTGAAGAGATTGA-3’, sgRNA 2, 5’-TCGGTAACTGACTTGAATGTCCA-3’), against mouse ICAM-1 (sgRNA 1, 5’-GAAGGCTTCTCTGGGATGGA-3’, sgRNA 2, 5’-GCAGGAAGGCTTCTCTGGGA-3’) (synthesized by Genewiz) were annealed and cloned into lentiCRISPR-v2-Puro vector (Addgene #52961) according to the published protocol (Sanjana et al., 2014). The constructed plasmids were packaged into lentiviral particles using 293T cells. WM9 and YUMM1.7 cells were infected with lentivirus and selected by 2 μg/ml puromycin for 2 days. Single cell clones were isolated using limited dilution. Knockout clones were identified by flow cytometry analysis and western blotting.

Purification of EVs

For collection of EVs, cells were cultured in media supplemented with 10% exosome-depleted FBS, in which EVs were depleted by overnight centrifugation at 100,000 g. Supernatants were then collected 48-72 h later for EV purification. Briefly, culture supernatants were centrifuged at 2,000 g for 20 min at 4°C to remove cell debris and dead cells (Beckman Coulter, Allegra X-14R). Supernatants were obtained and microvesicles were pelleted after centrifugation at 16,500 g for 45 min at 4°C (Beckman Coulter, J2-HS) with PBS. Supernatants were further centrifuged at 100,000 g for 2 h at 4°C (Beckman Coulter, Optima XPN-100. The pelleted membranes were suspended in PBS, and the exosomes were further purified by ultracentrifugation at 100,000 g for 2 h.

Nanoparticle tracking analysis

For exosomes purified by differential ultracentrifugation, the harvested exosomes were diluted 1:1000 with filtered PBS and loaded with a 1ml clean syringe to measure the size and concentration of the exosomes using a NanoSight NS300 (Malvern Instruments). For cell culture supernatant, the culture medium was collected and centrifugated at 2,000 g for 20 min at 4°C (Beckman Coulter, Allegra X-14R) to remove the cell debris and dead cells, and centrifugated at 16,500 g for 45 min at 4°C (Beckman Coulter, J2-HS) to remove microvesicles if necessary. The supernatants were carefully harvested for measurement by nanoparticle tracking analysis. To examine exosome secretion from Hrs or Rab27A KD cells, the cells were seeded at 2 × 10⁵ cells/well in 6-well plate overnight and then washed with PBS followed by 2 ml Opti-MEM™ (Thermo Fisher Scientific). After 8 hours, the media were collected for nanoparticle tracking analysis as described above.

Iodixanol density gradient centrifugation

For iodixanol density gradient centrifugation, a discontinuous iodixanol gradient (5%, 10%, 20% and 40%) was made by diluting 60% OptiPrep aqueous iodixanol with 0.25 M sucrose in 10 mM Tris. Exosomes purified by differential centrifugation were further loaded on the top of the discontinuous iodixanol gradient and then centrifuged at 100,000 g for 18 h at 4°C (Beckman Coulter, Optima MAX-XP). With the exosomes distributed at the density ranging from 1.13 to 1.19 g/ml, twelve fractions with equal volume were collected from the top of the gradients. The exosomes were finally pelleted by ultracentrifugation at 100,000 g for 2 h at 4°C and analyzed with western blotting.

Immune electron microscopy

For double immunogold labeling, exosomes in PBS were placed on formvar carbon-coated nickel grids, blocked, and incubated with rabbit anti-human monoclonal antibody that recognizes the extracellular domain of ICAM-1 (clone E3Q9N, Cell Signaling Technology) and mouse anti-human monoclonal antibody that recognizes the extracellular domain of PD-L1 (clone 5H1-A3, gift from Dr. Haidong Dong, Mayo Clinic) (Dong et al., 2002). Exosomes were then incubated with an anti-rabbit secondary antibody conjugated with protein A-gold particles (10 nm) and an anti-mouse secondary antibody conjugated with protein A-gold particles (5 nm). Each step was followed by four times PBS wash and 10 times ddH₂O wash before being contrast stained with 2% uranyl acetate. After air-drying, grids were visualized using a JEM-1011 transmission electron microscope.

Western blotting

Western blotting was performed according to previous protocol (Chen et al., 2018). In brief, cells or exosomes were lysed, and proteins were separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The samples were further electroblotted onto polyvinylidene fluoride membranes. After being blocked with 5% skimmed milk at room temperature for 1 h, the blots were incubated with primary antibodies overnight at 4°C. The blots were then probed with horseradish peroxidase-conjugated secondary antibodies for 1 hr at room temperature. The blots were subsequently developed with ECL Western Blotting Substrate (Pierce). Protein bands from western blotting were quantified with Fiji/Image J (Schindelin et al., 2012).

Fluorescence microscopy

Cells cultured on glass coverslips were fixed in 4% paraformaldehyde for 15 min, then permeabilized within 0.1% Triton X-100 in PBS for 20 min. After incubation with primary antibodies overnight at 4°C, the cells were incubated with fluorescence labeled secondary antibodies (Life Technologies) for 1 h. Stained samples were mounted with ProLong® Gold Antifade Reagent with DAPI (#8961, Cell Signaling Technology). Samples were observed using an Eclipse TE2000-U inverted microscope (Nikon) equipped with a PLAN APO ×100 1.3 NA objective and Cascade 512B CCD camera (Photometries) driven by Metamorph imaging software (Molecular Devices). The images were analyzed using NIS-Elements Advanced Research software (Nikon; version 4.50).

Immunoprecipitation

To analyze the mechanism of ICAM-I loading onto exosomes, 293T cells were co-transfected with Flag-Hrs and ICAM-1 plasmids before being lysed. The cleared lysate was incubated with Anti-FLAG Affinity Gel (Sigma-Aldrich) overnight at 4°C. The immunoprecipitated proteins were resolved by SDS-PAGE and western blotting.

Enzyme-linked immunosorbent assay (ELISA)

To detect ICAM-1 on extracellular vesicles in cell supernatants or patient plasma, 96-well ELISA plates (Biolegend) were coated with 0.5 μg per well (100 μl) of monoclonal antibody against ICAM-1 (clone HA58, Biolegend) overnight at 4°C. After blocking free binding sites with 200 μl of blocking buffer (Biolegend) for 1 h at room temperature, 100 μl EV samples purified from cell culture supernatants were added to each well, and incubated overnight at 4°C. To determine levels of ICAM-1, EVs purified from cell culture supernatants were prepared by serial dilution according to the total protein level. The level of ICAM-1 on the surface of exosomes derived from indicated cells was calculated based on the linear range of the ELISA. Afterwards, 100 μl biotinylated monoclonal ICAM-1 antibody (clone HCD54, Biolegend) was added to each well and incubated for 1 h at room temperature. A total of 100 μl horseradish peroxidase-conjugated streptavidin (BD Biosciences) diluted in PBS containing 0.1% BSA was added into each well and incubated for 1 h at room temperature. The ELISA plates were developed using tetramethylbenzidine (Pierce) and stopped with 0.5N H₂SO₄ before reading at 450 nm using a BioTek plate reader. Recombinant human ICAM-1 protein (R&D Systems, Cat# ADP4-050) was used to set a standard curve.

To detect the ICAM-1 level on PD-L1⁺ exosomes in healthy donor and melanoma patient plasma, exosomes isolated from plasma samples were dissolved in the same volume of PBS as the original plasma. The 96-well ELISA plates were coated with antibodies against PD-L1 (clone 5H1-A3, Mayo Clinic) as described above, and 100 μl of exosome samples purified from plasma were added to each well and incubated overnight at 4°C. Afterwards, 100 μl biotinylated monoclonal ICAM-1 antibody (clone HCD54, Biolegend) was added to each well and incubated for 1 h at room temperature, followed by incubation with 100 μl horseradish peroxidase-conjugated streptavidin for 1 h at room temperature. The ELISA plates were developed and stopped with 0.5N H2SO4 and read at 450 nm using a BioTek plate reader. Measurement of cytokines (IFN-γ, TNF-α and IL-2) secreted from human CD8⁺ T cells was performed using commercial kits according to manufacturer’s instructions (Biolegend).

ExoView single vesicle assay

Exosomes isolated from WM9 cell culture were run on a custom ExoView chip with capture antibody against PD-L1 (MIH3) (NanoView Biosciences Inc.). ExoView assay was run according to the standard protocol provided by NanoView Biosciences, Inc. Briefly, samples were diluted four times in incubation buffer, then 35 μL sample was incubated on the ExoView chip overnight (16 hours) at room temperature. After sample incubation chips were washed and stained for 1 h with an antibody cocktail that consist of pan-Tetraspanin (anti-CD81(JS-81), anti-CD63 (H5C6), and anti-CD9 (HI9a) CF488, anti-PD-L1 (MIH3) CF555, and anti-ICAM (HA58) CF647. Tetraspanin antibodies were sourced from NanoView Biosciences, Inc. PD-L1 and ICAM antibodies were purchased from Biolegend Inc. After staining, the chips were washed, dried and imaged with the R100 reader. NanoViewer Analysis software 3.04 was used to calculate the particle count and colocalization for each capture spot.

Exosomes sub-population removal

500 μl of magnetic beads (MagniSort™ Streptavidin Positive Selection Beads, Invitrogen, Catalog Number: MSPB-6003) were washed three times using PBS, and then incubated with biotinylated anti-ICAM-1 antibodies (1:50, clone HCD54, Biolegend) or, as a control, biotinylated anti-human IgG Fc antibodies (1:50, clone HP6017, Biolegend) on a shaker at room temperature for 1 h. The beads were then placed on the magnet for 1 min and washed with PBS. 1 mg of exosomes (in 1 ml PBS) was used in the experiment, of which 500 μg was incubated with biotinylated anti-IgG antibody or ICAM-1 antibody-coupled magnetic beads at 4°C overnight. After magnet absorption, the supernatant was transferred to a new tube for ultracentrifugation to obtain the exosomes after depletion of ICAM⁺ exosomes as the void exosome group. The magnetic beads bound with ICAM⁺ exosomes were collected and washed 3 times with PBS to obtain purified ICAM⁺ exosomes. The same amounts of proteins (10 μg) for each group were loaded for western blot analysis.

Treatment of CD8⁺ T cells with the exosomes

To block exosomal ICAM-1, 200 μg of purified exosomes were incubated with 10 μg/ml ICAM-1 blocking antibodies or IgG isotype antibodies in 100 μl PBS, washed with 30 ml PBS. To remove the free antibodies, exosomes were washed with PBS and collected through ultracentrifugation. Human CD8⁺ T cells were obtained from the Human Immunology Core at the University of Pennsylvania Perelman School of Medicine and only used when the purity of CD8⁺ T cells was >90%. Mouse CD8⁺ T cells were purified from splenocytes using the Dynabeads Untouched Mouse CD8 Cells Kit (Invitrogen). The purity of splenic CD8⁺ T cells was >90% in most purifications. The CD8⁺ T cells were stimulated with 2 μg/ml anti-CD3 (clone OKT3, Bio X Cell) and anti-CD28 (clone CD28.2, BD Biosciences) antibodies for 24 h and then incubated with human melanoma cell-derived exosomes or mouse YUMM1.7 cell-derived exosomes with or without ICAM-1 blocking for 48 h. For human CD8⁺ T cells (2 × 10⁵ cells/well in a 96-well plate), 25 μg/ml of human WM9 cell-derived exosomes were used according to our previous study (Li et al., 2019). For mouse CD8⁺ T cells (2 × 10⁵ cells/well in a 96-well plate), 25 μg/ml of YUMM1.7 cell derived exosomes were used. The treated cells were analyzed using flow cytometry. Human H1264, H1299, HCT116 and murine MC38 cell-derived exosomes were also used in this study. Information for the primary antibodies is included in Supplemental Table 1.

Exosome-T cell binding assay

To verify the role of exosomal ICAM-1 in the interactions between melanoma cell-derived exosomes and CD8⁺ T cells, exosomes with or without ICAM-1 blocking antibodies or IgG isotype antibodies were stained with CFSE in 100 μl PBS, washed with 10 ml PBS, and pelleted by ultracentrifugation. Unstimulated or stimulated human CD8⁺ T cells (2 × 10⁵ cells/well in 96-well plates) were incubated with 25 μg/ml CFSE-labeled exosomes for 24 h, and then prepared for flow cytometry analysis. Exosomes purified from WM9 cells with scrambled shRNA or shRNAs against human ICAM-1 were also used for the exosome-T cell binding assay.

Flow cytometry

The human peripheral blood mononuclear cells (PBMCs) and CD8⁺ T cells were obtained from Human Immunology Core at the University of Pennsylvania Perelman School of Medicine. After indicated treatment, fresh PBMC samples or CD8⁺ T cells were analyzed using flow cytometry as previously described (Chen et al., 2018). In brief, live or dead cells were discriminated using the Live/Dead Fixable Aqua Dead Cell Stain Kit (Life Technologies). Then the cells were blocked and cellular surface staining was performed for 30 min at 4°C. After using a fixation/permeabilization kit (eBioscience), intracellular staining was conducted for 60 min on ice. The samples were washed with PBS containing 0.2% BSA and analyzed using BD™ LSR II flow cytometer. The results were analyzed by FlowJo v10.0. Information about the primary antibodies for the flow cytometry is included in Supplemental Table 2.

The ETIDS system

The SirA sequence, including a terminal Flag-tag, was cut from the SrtA-PD-1 plasmid (gift from Gabriel D. Victora, Rockefeller University) (Pasqual et al., 2018) and ligated with PD-L1 (PD-L1-SrtA). A five-glycine tag (G5) followed by a Myc tag was cut and fused to the N-terminus of PD-1 (PD-1-G5). The sequences of all constructs were verified by sequencing. PD-L1-SrtA and PD-1-G5 plasmids were transfected into WM9 and Jurkat cells, respectively, using Lipofectamine 2000 (Thermo Fisher Scientific). The tdTomato (PD-L1-SrtA) and GFP (PD-1-G5)-expressing cells were enriched by fluorescence-activated cell sorting. Protein expression was then verified by western blotting.

For ETIDS in vitro labelling experiments, stable transfected Jurkat cells were detached using a non-enzymatic cell dissociation solution (Thermo Fisher Scientific), washed and resuspended at 7.5 × 10⁶ cell per ml in PBS. 30 μg of exosomes derived from WM9-PD-L1-SrtA were added into 200 μl above cell suspension (1.5 × 10⁶ Jurkat cells), to which biotin-LPETG was added at a final concentration of 100 μM. Cells were incubated at 37°C for 4 h and washed three times with PBE (PBS+0.2%BSA+10 mM EDTA) buffer to remove excess attached exosomes. Streptavidin-Alexa fluor 647 was incubated for 1 h at 4 °C before FACS analysis.

Animal studies

To establish the syngeneic mouse melanoma model, YUMM1.7 cells (1 × 10⁶ cells in 100 μl medium) were subcutaneously injected into immunocompetent C57BL/6 mice (female, 6 to 8 weeks). Five days after implantation, the exosomes derived from control or ICAM-1 KO YUMM1.7 cells, or the YUMM1.7 cell-derived exosomes blocked with the IgG isotype antibodies, anti-ICAM-1 antibodies (10 μg/ml) were injected. Tail vein injections of exosomes (50 μg in 100 μl PBS) were performed every 3 days and the mice were weighted. Tumors were measured using a digital caliper, and tumor volume was calculated by the formula: length × (width)²/2. The mice were euthanized before the longest dimension of the tumors reached 2.0 cm. Mice were allocated randomly to each treatment group. For flow cytometry, single cell suspensions of tumor cells were prepared and red blood cells were lysed using ACK Lysis Buffer. Flow cytometry was performed in a double blind fashion. Briefly, tumor samples were harvested and single cell suspensions were prepared. Red blood cells were lysed using ACK Lysis Buffer. Live or dead cells were discriminated using Live/Dead Fixable Aqua Dead Cell Stain Kit (Life Technologies). Cell suspensions were then blocked and cell surface staining was performed for 25 min at 4°C. Intracellular staining was performed for 60 min on ice after using a fixation/permeabilization kit (eBioscience). Information about the primary antibodies is included in Key Resources Table. MC38 cells were also used to establish the syngeneic mouse colon cancer model following the same procedures.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-human ICAM-1	Cell Signaling Technology	Cat# 67836; RRID:AB_2799738
Mouse monoclonal anti-human ICAM-1	BioLegend	Cat# 353102; RRID:AB_11204426
Mouse monoclonal anti-human ICAM-1, biotin	BioLegend	Cat# 322706; RRID:AB_535978
Mouse monoclonal anti-human ICAM-1	BioLegend	Cat# 322702; RRID:AB_535974
Mouse monoclonal anti-human PD-L1	Haidong Dong, Mayo Clinic	PMID: 21355078
Rabbit monoclonal anti-human PD-L1	Cell Signaling Technology	Cat# 86744; RRID:AB_2800088
Mouse monoclonal anti-human PD-L1, biotin	eBioscience	Cat# 13-5983-82; RRID:AB_466840
Mouse monoclonal anti-human PD-1	BioLegend	Cat# 329912; RRID:AB_961417
Mouse IgG isotype control	BioLegend	Cat# 401404; RRID:AB_2801451
Rat monoclonal anti-mouse PD-L1	BioXCell	Cat# BE0101; RRID:AB_10949073
Rat IgG isotype control	BioXCell	Cat# BE0090; RRID:AB_1107780
Rat monoclonal anti-mouse CD16/32	BioLegend	Cat# 101302; RRID:AB_312801
Mouse monoclonal anti-human ICAM-1	BioLegend	Cat# 353113; RRID:AB_2715941 Cat# 353106; RRID:AB_10897647
Mouse monoclonal anti-human PD-L1	BioLegend	Cat# 329706; RRID:AB_940368
Mouse monoclonal anti-human PD-L1	BD Biosciences	Cat# 558065; RRID:AB_647176
Mouse monoclonal anti-human CD8a	eBioscience	Cat# 48-0088-42; RRID:AB_1272062
Mouse monoclonal anti-human CD4	Biolegend	Cat# 317416; RRID:AB_571945
Mouse monoclonal anti-human PD-1	BioLegend	Cat# 329904; RRID:AB_940479
Mouse monoclonal anti-human Ki-67	BD Biosciences	Cat# 561283; RRID:AB_10716060
Mouse monoclonal anti-human Granzyme B	Life Technologies	Cat# GRB04; RRID:AB_2536538
Rat monoclonal anti-mouse PD-1	BioLegend	Cat# 109110; RRID:AB_572017
Rat monoclonal anti-mouse Ki-67	BioLegend	Cat# 652420; RRID:AB_572017
Rat monoclonal anti-mouse Granzyme B	eBioscience	Cat# 12-8898-82; RRID:AB_10870787
Rat monoclonal anti-mouse CD45	BioLegend	Cat# 157204; RRID:AB_2876533
Rat monoclonal anti-mouse CD3	BioLegend	Cat# 100204; RRID:AB_312661
Rat monoclonal anti-mouse CD8b	BioLegend	Cat# 126610; RRID:AB_2260149
Rat monoclonal anti-mouse CD4	BioLegend	Cat# 100451; RRID:AB_2564591
Mouse monoclonal anti-human CD11a	Biolegend	Cat# 301341; RRID:AB_2563371
Mouse monoclonal anti-human CD11b	Biolegend	Cat# 301309; RRID:AB_314161
Mouse monoclonal anti-human CD11c	Biolegend	Cat# 301604; RRID:AB_314174
Mouse monoclonal anti-human LFA-1	Biolegend	Cat# 363410; RRID:AB_2716070
Mouse monoclonal anti-human CD18	Biolegend	Cat# 373406; RRID:AB_2716022
Mouse monoclonal anti-human CD11a	Biolegend	Cat# 301213; RRID:AB_314151
Mouse monoclonal anti-human CD11b	Biolegend	Cat#101213; RRID:AB_312796
Mouse monoclonal anti-human CD18	Biolegend	Cat# 302102; RRID:AB_314220
Mouse monoclonal anti-human CD63	Abcam	Cat# ab8219; RRID:AB_306364
Mouse monoclonal anti-human CD63	Biolegend	Cat# 353039; AB_2800940
Rat monoclonal anti-mouse CD63	Biolegend	Cat# 143901; RRID:AB_11203908
Mouse monoclonal anti-human CD8α	Biolegend	Cat# 372902; RRID:AB_2650657
Rabbit monoclonal anti-HRS	Cell Signaling Technology	Cat# 15087S; RRID:AB_2798700
Rabbit anti-human CD63	Abcam	Cat# ab68418; AB_10563972
Rabbit monoclonal anti-TSG101	Abcam	Cat# ab125011; RRID:AB_10974262
Rabbit monoclonal anti-human PD-L1	Cell Signaling Technology	Cat# 13684S; RRID:AB_2687655
Rabbit monoclonal anti-mouse PD-L1	Abcam	Cat# ab213480; RRID:AB_2773715
Rabbit monoclonal anti-Alix	Cell Signaling Technology	Cat# 2171S; RRID:AB_2299455
Rabbit monoclonal anti-human CD9	Cell Signaling Technology	Cat# 13174S; RRID:AB_2798139
Rabbit monoclonal anti-GAPDH	Cell Signaling Technology	Cat# 5174S; RRID:AB_10622025
Rabbit monoclonal anti- DYKDDDDK (Flag)	Cell Signaling Technology	Cat# 14793S; RRID:AB_2572291
Rabbit monoclonal anti-mouse CD8α	Cell Signaling Technology	Cat# 98941; RRID:AB_2756376
Biotin donkey anti-rabbit IgG	Jackson Immuno Research	Cat# 711-065-152; RRID:AB_2340593
Bacterial and virus strains
Stellar™ Competent Cells	Takara	Cat#636766
Lentivirus	This paper	N/A
Biological samples
Melanoma patient peripheral blood samples	University of Pennsylvania Abramson Cancer Center	UPCC# 08607 / IRB# 703001
Healthy donor peripheral blood samples	The Wistar Institute	N/A
PBMCs	Penn Human Immunology Core	N/A
CD8 cells	Penn Human Immunology Core	N/A
Chemicals, peptides, and recombinant proteins
Recombinant Human IFN-γ	PeproTech	Cat# 300-02
Recombinant Murine IFN-γ	PeproTech	Cat# 315-05
Biotin-Ahx-LPETGS-NH2	Lifetein	Cat# 5467
Critical commercial assays
Dynabeads Untouched Mouse CD8 Cells Kit	Invitrogen	Cat# 11417D
MagniSort™ Streptavidin Positive Selection Beads	Invitrogen	Cat# MSPB-6003
Anti-FLAG Affinity Gel	Sigma-Aldrich	Cat# A2220; RRID:AB_10063035
PD-1/PD-L1 Blockade Bioassay	Promega Corporation	Cat# J1250
MagniSort™ Streptavidin Positive Selection Beads	Invitrogen	Cat# MSPB-6003
DAB Substrate Kit	Fisher Scientific	Cat# BD 550880; RRID:AB_2868905
Experimental models: Cell lines
WM9	Meenhard Herlyn, The Wistar Institute	PMID: 30089911
WM164	Meenhard Herlyn, The Wistar Institute	PMID: 30089911
YUMM1.7	Marcus Bosenberg, Yale School of Medicine	PMID: 28379630
H1264	Steven Albelda, Penn Perelman School of Medicine	PMID: 19671764
H1299	Steven Albelda, Penn Perelman School of Medicine	PMID: 19671764
HCT116	Shiaw-Yih Lin, MD Anderson Cancer Center	PMID: 32109374
MC38	Serge Y. Fuchs, Penn Veterinary School of Medicine	PMID: 32807917
HEK293T	ATCC	CRL-3216
WM9 PD-L1 KO	This paper	N/A
YUMM1.7 ICAM-1 KO	This paper	N/A
Experimental models: Organisms/strains
C57BL/6	The Jackson Laboratory	Cat#000664
Oligonucleotides
Human PD-L1 KO sgRNA 1: CCTTGCACTTCTGAAGAGATTGA	This paper	N/A
Human PD-L1 KO sgRNA 2: TCGGTAACTGACTTGAATGTCCA	This paper	N/A
Mouse ICAM-1 KO sgRNA 1: GAAGGCTTCTCTGGGATGGA	This paper	N/A
Mouse ICAM-1 KO sgRNA 2: GCAGGAAGGCTTCTCTGGGA	This paper	N/A
Human ICAM-1 shRNA 1: GCCAACCAATGTGCTATTCAA	The Wistar Institute	TRCN0000029630
Human ICAM-1 shRNA 2: CCTCAGCACGTACCTCTATAA	The Wistar Institute	TRCN0000029631
Human HRS shRNA 1: GCACGTCTTTCCAGAATTCAA	The Wistar Institute	TRCN0000037898
Human HRS shRNA 2: GCATGAAGAGTAACCACAGC	Alissa Weaver, Vanderbilt University School of Medicine	PMID: 25968605
Human RAB27A shRNA 1: GCTGCCAATGGGACAAACATA	Alissa Weave, Vanderbilt University School of Medicine	PMID: 25968605; TRCN0000005297
Human RAB27A shRNA 2: CAGGAGAGGTTTCGTAGCTTA	The Wistar Institute	TRCN0000005298
Recombinant DNA
Plasmid: PD-L1-SrtA	This paper	N/A
Plasmid: PD-1-G5	This paper	N/A
Plasmid: Flag-HRS	This paper	N/A
Plasmid: pCDM8-hICAM-1	Addgene	Cat#8632
Software and algorithms
GraphPad software	Prism 8.0	https://www.graphpad.com/scientificsoftware/prism/
Fiji	(Schindelin et al., 2012)	https://imagej.net/Fiji
FlowJo	Version v10	https://www.flowjo.com/solutions/flowjo/downloads

Immunohistochemistry

The MC38 xenograft tumors were harvested and formalin-fixed. The samples were embedded, sectioned and IHC staining was performed at The Wistar Institute Histotechnology Facility. Antigen retrieval was performed by steaming the slides in citrate buffer (pH 6.0) for 5 min. After washing with PBS, the sections were blocked and incubated with an anti-CD8α (1:200; Cell Signaling) antibody overnight at 4°C, followed by incubation with a biotinylated secondary antibody (1:200; Jackson Immuno Research) for 30 min. Detection was performed using DAB (Fisher Scientific). The slides were scanned in iHisto Inc. and visualized using QuPath-0.2.3 software. The CD8⁺ T cells that stained with strong membranous positivity were enumerated in five separate areas at 20× magnification in a blinded fashion.

QUANTIFICATION AND STATISTICAL ANALYSIS

All the statistical analyses were performed using GraphPad Prism v.8.0 software. Normality of distribution was determined by D’Agostino-Pearson omnibus normality test and the equal variance assumption among between groups was assessed by Brown-Forsythe test. For equal variance data, significance of mean differences was determined using unpaired Student’s t-test (two groups) or one-way ANOVA with appropriate post-hoc tests (more than two groups); for groups that differed in variance, unpaired t-test with Welch’s correction (two groups) or Welch’s ANOVA with appropriate post-hoc tests (more than two groups) was performed. Two-way ANOVA was used to compare mouse volume data among different groups. Error bars shown in graphical data represent mean ± s.d. A two-tailed value of P < 0.05 was considered statistically significant.

Highlights.

• ICAM-1 and PD-L1 co-localize on exosomes and are both upregulated by interferon-γ.

• ICAM-1 is a prerequisite for exosomal PD-L1 mediated inhibition of CD8⁺ T cells.

• Reciprocal upregulation of ICAM-1 and LFA-1 promotes exosome-T cell interaction.

• Established an assay system for the interaction of exosomal PD-L1 with PD-1 on T cells.