Extracellular vesicle surface display enhances the therapeutic efficacy and safety profile of cancer immunotherapy

Abstract

Immunotherapy has emerged as a mainstay in cancer therapy, yet its efficacy is constrained by the risk of immune-related adverse events. In this study, we present a nanoparticle-based delivery system that enhances the therapeutic efficacy of immunomodulatory ligands while concurrently limiting systemic toxicity. We demonstrate that extracellular vesicles (EVs), lipid bilayer enclosed particles released by cells, can be efficiently engineered via inverse electron demand Diels-Alder (iEDDA)-mediated conjugation to display multiple immunomodulatory ligands on their surface. Display of immunomodulatory ligands on the EV surface conferred substantial enhancements in signaling efficacy, particularly for tumor necrosis factor receptor superfamily (TNFRSF) agonists, where the EV surface display served as an alternative FcγR-independent approach to induce ligand multimerization and efficient receptor crosslinking. EVs displaying a complementary combination of immunotherapeutic ligands were able to shift the tumor immune milieu toward an anti-tumorigenic phenotype and significantly suppress tumor burden and increase survival in multiple models of metastatic cancer to a greater extent than an equivalent dose of free ligands. In summary, we present an EV-based delivery platform for cancer immunotherapeutic ligands that facilitates superior anti-tumor responses at significantly lower doses with fewer side effects than is possible with conventional delivery approaches.

Keywords: cancer immunotherapy; extracellular vesicles; iEDDA; nanomedicine; surface functionalization.

Graphical abstract

Figure 1

iEDDA-mediated conjugation facilitates efficient EV surface functionalization (A) Schematic depicting iEDDA-mediated EV surface functionalization. (B) Single EV flow cytometric analysis assessing the conjugation efficiency of varying concentrations of fluorescent AZ488-TCO with MTet-EVs. The concentration of AZ488-TCO used is indicated in the top left corner of the histogram. (C) Single EV flow cytometric analysis assessing Biotin-TCO conjugation on MTet-EVs. EVs were co-stained with anti-GPA antibody to accurately gate RBCEVs. Percentages on the gated plot are shown for MTet-EV + Biotin-TCO. (D) Quantification of Biotin-TCO probes conjugated per EV determined using competition ELISA. (E) Copy number of IgG conjugated per EV, determined using a total Rat IgG ELISA kit. (F) Representative TEM micrographs of EVs before and after iEDDA-mediated conjugation. Scale bar, 300 nm. (G) Size distribution profiles of unmodified and iEDDA-conjugated EVs determined using NTA. The mean diameter is noted on each histogram. (H) Single EV flow cytometric analysis of MTet-EVs conjugated with pairs of TCO-labeled proteins. Figures (D) and (E) represent data from three individual replicates prepared from separate batches of EVs. For (D) and (E), the concentration in μM depicts the concentration of TCO-labeled biotin/IgG incubated with MTet-EVs for the iEDDA reaction. MTet: methyltetrazine, TCO: trans-cyclooctene, iEDDA: inverse electron demand Diels-Alder. In (C) molecular weights of protein markers in kDa are shown on the left. Error bars represent standard deviation.

Figure 2

EV-mediated surface display of ligands enhances signaling compared with equivalent doses of free ligands (A) Schematic depicting the differences in receptor engagement between free agonistic CD137 antibodies, or a similar number of antibodies conjugated on a single EV. (B) Immunofluorescent imaging of T cells incubated with EVs conjugated with an isotype control antibody (EV-IgG) or an agonistic CD137 antibody (EV-CD137 Ab). T cells were stained for CD45 (green) to visualize the cell membrane while EVs were visualized using CellTrace Yellow labeling (red). Cells were counterstained with Hoechst (cyan). Scale bar, 50 μm. (C) EV-cell association assay comparing the relative binding affinity of CD137 antibody-conjugated EVs (EV-CD137 Ab) with EVs conjugated with an isotype control antibody (EV-IgG). EV binding was quantified using flow cytometry staining for GPA, an RBCEV-specific surface marker. (D) Mean fluorescent intensity (MFI) of GPA staining from (C). (E) IFN-γ ELISA for supernatants of T cells treated with equivalent quantities of EV-conjugated, liposome-conjugated or free CD137 agonistic antibody. (F) Effect of increasing doses of CD137 agonistic antibody-conjugated EVs on IFN-γ release. The numbers in μM depict the concentrations of TCO-labeled CD137 antibody incubated with MTet-EVs for the iEDDA reaction. (G) IFN-γ released from activated T cells following stimulation with free (soluble) mouse CD137L or EV-conjugated CD137L. (H) Relative fold change in T cell proliferation in the presence of free OX40 Ab or an equivalent dose of EV-conjugated OX40 Ab quantified using alamarBlue assay. T cells were activated with CD3/CD28 beads prior to OX40 stimulation. (I) Relative fold change in T cell proliferation compared with an untreated control at varying doses of free recombinant mouse IL-2 or EV-conjugated IL-2. (J and K) MFI of CD69 (J) and CD25 (K) on T cells over a period of 5 days following treatment with a cocktail of agonistic CD3/CD28/CD137 antibodies in their free form or conjugated on EVs in cis/trans orientation. A positive control (Beads-Cis) of Dynabeads conjugated with the same ligands in a similar orientation as shown in EV-Cis is included. (L) Proliferation of T cells following stimulation with each of the indicated treatments, using alamarBlue assay. For (J)–(L), the asterisks are shown to denote significant differences between EV-Cis and EV-Trans. For (F)–(L), data were obtained from three to four independent repeats performed using EVs from individual donors and T cells from different mice. MTet: methyltetrazine; GPA: glycophorin A; iEDDA: inverse electron demand Diels-Alder; DOPE: dioleoylphosphatidylethanolamine; PS: phosphatidylserine. Student’s two-tailed t test (D) and (E), (G)–(I), two-way ANOVA test (F), (J)–(L): ns, not significant, ∗p < 0.05, ∗∗∗p < 0.001. Error bars represent standard deviation.

Figure 3

Locally administered EVs are well retained in the immediate tumor microenvironment (A) Illustration of intratracheal administration in mouse models. (B) Representative near-IR fluorescent images of Free Ligand or EV-Ligand labels in organs acquired using IVIS 4 h post-administration of a single intratracheal dose into C57BL/6 mice. EV-Ligand or free ligand proteins were labeled with CF Dye TFP Ester CF750, a near-IR amine-reactive dye prior to administration to facilitate tracking of in vivo biodistribution. A flowthrough control is included for reference. (C) Summary of the in vivo biodistribution of Free Ligand and EV-Ligand treatments, represented as the relative percentage of the total administered dose present in each organ (n = 3 mice per condition). (D) Relative accumulation of Free Ligand and EV-Ligand treatments in the serum of the mice over a period of 42 h, monitored via measurement of near-IR signals in the serum collected at each time interval. (E) Blood chemistry analysis of mice administered with six doses of free ligand or an equivalent dose of EV-conjugated ligands, assessed 20 days after the first administration. Mice were administered the treatments intratracheally at 3-day intervals. (F) Cellular uptake of control (EV/EV-IgG) or T cell-targeted EV-Ligand treatments by immune cell subsets in the lung assessed 5 h after administration of a single dose of EVs. For (C)–(F), data were obtained from three to four independent repeats performed using EVs from individual donors. The ligand combination used in panels (B)–(F) consisted of αCD3, αCD137, αPD-1, and mIL-2. Student’s two-tailed t test (E) and (F), two-way ANOVA test (D): ns, not significant, ∗p < 0.05, ∗∗∗p < 0.001. Error bars represent standard deviation.

Figure 4

T cell-targeted immunomodulatory EVs improve tumor suppression in a mouse model of lung metastatic melanoma (A) Outline of the in vivo experiment utilized to assess the relative efficacy of EV-Ligand treatments compared with equivalent doses of free ligands. A luciferase reporter was used to monitor the implantation of B16-F10-Luc2 cells and tumor progression in the lung. (B) Luciferase signals from the lungs of mice administered with each of the indicated treatments over a 20-day period, quantified using IVIS. (C and D) AST (C) and ALT (D) levels in the serum of mice on day 20. (E) Ex vivo tumor cell killing assay performed using T cells isolated from mouse lungs at the end of the 20-day treatment period. (F) Immunofluorescent staining of lung sections from each treatment condition, assessing T cell infiltration and tumor burden. A cocktail of melanoma-specific markers that are overexpressed in B16 F10-Luc2 cells was used to detect tumor cells (pan-cytokeratin, podoplanin, PD-L1 and firefly luciferase). Scale bar, 50 μm. (G) Quantification of T cells in the lung across each of the conditions in (F). Quantification is based on 20 images from each treatment condition, acquired in a double-blinded manner. (H) Quantification of tumor burden based on 20 tile scans of parasagittal lung sections from each treatment condition. (I) Representative H&E images of lung sections from each treatment condition. Scale bar, 250 μm. (J) Body weight of mice from each treatment group monitored over an extended duration until mice succumbed to the tumor or reached the conditions for symptom-free survival. (K) Survival of mice administered with each treatment over a period of 38 days. (B), (J), and (K) include data from five to seven mice per treatment condition. (C)–(E) include data from five to six mice for each condition. Images in (F) were acquired in a blinded manner. Two-way ANOVA test (B), (J), Log rank (Mantel-Cox) test (K), and Student’s two-tailed t test (C)–(E), (G), and (H): ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Error bars represent standard deviation.

Figure 5

Surface display of ligands synergizes with intraluminal loading of innate immune agonists (A) IL-12p70 levels in cell culture supernatants determined using ELISA following treatment of bone marrow-derived macrophages (BMDMs) or bone marrow-derived dendritic cells (BMDCs) with free murine CD40 agonistic antibody or equivalent doses of EV-conjugated CD40 antibody. (B) MFI of MHC II and co-stimulatory molecules on the surface of BMDCs following CD40 stimulation as in (A). (C) IL-12p70 release in human monocyte-derived macrophages following stimulation with free agonistic human CD40 antibody or an equivalent dose of EV-conjugated antibodies. (D) Immunofluorescent imaging of BMDMs incubated with control EVs, CD40-conjugated EVs or an equivalent dose of free CD40 antibody. EVs were visualized using CellTrace Yellow labeling (red) and cells were counterstained using Hoechst (cyan). Free CD40 antibody was visualized using a secondary anti-rat IgG (magenta). Cells were stained with a separate CD40 antibody that detected the cytoplasmic domain of CD40 (green). Scale bar, 20 μm. (E and F) MFI of MHC II, CD80, and CD86 on the surface of BMDMs (E) or BMDCs (F) following stimulation with a fixed dose of free R848 or EV-loaded R848. EV-DMSO is included as a vehicle control. (G) IL12p70 release by BMDMs or BMDCs quantified using ELISA after 24 h stimulation with R848 or EV-R848. (H) Fold change in IL-12p70 expression following treatment with either CD40-conjugated EVs, R848-loaded EVs, or CD40-conjugated R848-loaded EVs. A co-treatment with equivalent doses of free CD40 antibody and R848 is included for comparison. For (A)–(C) and (E)–(H), data were obtained from three to four independent repeats performed using EVs from individual donors and macrophages/dendritic cells from different mice. Student’s two-tailed t test: ns, not significant, ∗p < 0.05, ∗∗∗p < 0.001. Error bars represent standard deviation.

Figure 6

Multifunctional immunomodulatory EVs result in enhanced tumor suppression in an autochthonous cell-derived model of PDAC (A) Schematic of the in vivo experiment used to assess the anti-tumor efficacy of different treatments. Following intravenous injection of KPCY cells and subsequent intratracheal treatments, mice were euthanized on day 30 to determine tumor burden, measure toxicity parameters, and perform immunophenotyping or alternatively maintained until they reached the criteria for symptom-free survival. In a separate experiment, lung-metastasis-bearing mice were subcutaneously injected with 0.5 M KPCY cells to generate a secondary tumor challenge and the tumor burden was monitored for up to 20 days. (B) tSNE plots based on multicolor flow cytometric analysis illustrating the changes in immune cell composition in tumor-bearing mouse lungs on day 30. An opt-SNE algorithm was used for automated clustering and visualization of immune cell subsets and manually gated immune cell populations were overlaid on the tSNE plots after dimensionality reduction. Unidentified cell populations are left in gray color and the proportion of tumor cells were included in the plot for reference (black). (C) Percentage of tumor cells in mouse lungs in each treatment group at the endpoint (day 30). Tumor cells were gated using flow cytometry based on their expression of YFP. (D and E) AST (D) and ALT (E) levels in the serum of mice on day 30. (F) Body weight of mice monitored over an extended period of time until the mice reached the criteria for symptom-free survival. (G) Survival of mice administered with each treatment over a period of 60 days. (H) Volume of the flank tumor used to assess response to secondary tumor challenge. Mice were injected with tumor cells on day 15 and tumor volume was monitored for 20 days post-implantation as shown in (A). Each line indicates one mouse. (C)–(H) include data from five to seven mice per treatment condition. Two-way ANOVA test (F), mixed effects analysis (H), Log rank (Mantel-Cox) test (G), and Student’s two-tailed t test (C–E): ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Error bars represent standard deviation.