Biosynthetic neoantigen displayed on bacteria derived vesicles elicit systemic antitumour immunity

Abstract

Neoantigens derived from mutant proteins in tumour cells could elicit potent personalized anti-tumour immunity. Nevertheless, the layout of vaccine vehicle and synthesis of neoantigen are pivotal for stimulating robust response. The power of synthetic biology enables genetic programming bacteria to produce therapeutic agents under contol of the gene circuits. Herein, we genetically engineered bacteria to synthesize fusion neoantigens, and prepared bacteria derived vesicles (BDVs) presenting the neoantigens (BDVs-Neo) as personalized therapeutic vaccine to drive systemic antitumour response. BDVs-Neo and granulocyte-macrophage colony-stimulating factor (GM-CSF) were inoculated subcutaneously within hydrogel (Gel), whereas sustaining release of BDVs-Lipopolysaccharide (LPS) and GM-CSF recruited the dendritic cells (DCs). Virtually, Gel-BDVs-Neo combined with the programmed cell death protein 1 (PD-1) antibody intensively enhanced proliferation and activation of tumour-infiltrated T cells, as well as memory T cell clone expansion. Consequently, BDVs-Neo combining with checkpoint blockade therapy effectively prevented tumour relapse and metastasis.

Keywords: bacteria derived vesicles (BDVs); cancer vaccines; checkpoint blockade; immunotherapy; neoantigen.

SCHEME 1

Schematic of preparation of Gel‐GFP‐Mutation‐M33‐M47 BDVs combined with PD‐1 antibody for immunotherapy of melanoma. (a) Preparation process of Gel‐BDVs. (b) Effector T cell were actived by Gel‐GFP‐Mutation‐M33‐M47 BDVs to against melanoma

FIGURE 1

Characterizations of GFP‐Mutation‐M33‐M47 E. coli BDVs. (a) Establishment of E. coli stably expressing GFP (Scale bar: 2 μm). (b) Schematic diagram illustration the presentation of neoantigens using Lpp‐OmpA system on E. coli outer membrane. (c) The TEM images shown the size and morphology of GFP‐Mutation‐M33‐M47 E. coli (Scale bar: 2 μm). (d) The TEM images shown the size and morphology of GFP‐Mutation‐M33‐M47 E. coli BDVs (Scale bar: 100 nm). (e) The confocal image of GFP‐Mutation‐M33‐M47 E. coli BDVs (Scale bar: 5 μm). (f) The size distribution of GFP‐Mutation‐M33‐M47 E. coli BDVs measured by dynamic light scattering (DLS). (g) The zeta potential of the E. coli BDVs. Blank (BL21(DE3)plysS): non‐heat shock transformation control, worked as a negative control. (h) Western blot analysis indicated the expression of GFP on E. coli and BDVs. (i) Western blot analysis indicated the expression of GFP and ompF on E. coli, BDVs, OM and IM

FIGURE 2

Dendritic cell activity and maturation in vitro and in vivo. (a) Confocal 3D image of BMDCs uptake of GFP‐Mutation‐M33‐M47 E. coli BDVs in vitro (Scale bar: 10 μm). Green: BDVs. Red: cell membrane. Blue: cell nucleus. (b, c) Quantitative analysis of CD11c⁺ CD40⁺ cells (b) and mature DCs (c) from different treatment groups in vitro (gated on CD11c⁺ cells, n = 5). Error bar, mean ± s.d.. PBS (G1), Blank BDVs (G2), Normal‐M33‐M47 BDVs (G3), Mutation‐M33‐M47 BDVs (G4). (d, e) Representative flow cytometry data (d) and statistical data (e) to show CD11c⁺ CD40⁺ cells induced by different formulations of BDVs in vivo (in lymph nodes) on day 3 post‐ injection (gated on CD11c⁺ cells, n = 5). Cells were stained with anti‐CD11c‐APC, anti‐CD40‐PE antibodies (Biolegend). (f, g) Representative flow cytometry data (f) and statistical data (g) to show DC maturation induced by different formulations of BDVs in vivo (in lymph nodes) on day 3 post‐injection (gated on CD11c⁺ cells, n = 5). Cells were stained with anti‐CD11c‐APC, anti‐CD80‐FITC, anti‐CD86‐PE antibodies (Biolegend). Error bar, mean ± s.d.. (h) Immunofluorescence analysis of CD11c⁺ DCs in the skin after the treatments (Scale bar: 50 μm). Gel (#1), Gel‐GM‐CSF (#2), Gel‐Blank BDVs + GM‐CSF (#3), Gel‐Normal‐M33‐M47 BDVs + GM‐CSF (#4), Mutation‐M33‐M47 BDVs + GM‐CSF (#5). mean ± s.d., n = 5. NS: no significant, *P < 0.05, **P < 0.01, ***P < 0.001. One‐way ANOVA with Tukey post‐hoc tests (b, c, e, g)

FIGURE 3

Antitumour recurrence effect of BDVs‐triggered cancer immunotherapy in the surgical bed of B16F10‐luc model. (a) Schematic illustration of BDVs‐triggered cancer vaccine in an incomplete‐surgery B16F10‐luc tumour model. (b) In vivo bioluminescence images of the B16F10‐luc tumour‐bearing mice (n = 5). (c) Single tumour volume of each mouse in different groups. (d) Tumour growth curves in B16F10‐luc model (n = 7), Error bar, mean ± s.e.m.. (e) Tumour weights in different treatment groups (n = 7), Error bar, mean ± s.d.. (f) Survival curves in different groups (n = 10) (#1) Gel‐PBS, (#2) Gel‐Blank BDVs, (#3) Gel‐Normal‐M33‐M47 BDVs, (#4) Gel‐Mutation‐M33‐M47 BDVs, (#5) aPD‐1 (anti‐PD‐1 antibody), (#6) Gel‐Mutation‐M33‐M47 BDVs + aPD‐1. NS: no significant, *P < 0.05, **P < 0.01, ***P < 0.001. One‐way ANOVA with Tukey post‐hoc tests (d, e) or the Long‐Rank (Mantel‐Cox) test (f)

FIGURE 4

BDV‐neoantigen vaccine induced a robust antitumour immune response. (a, b) Representative flow cytometry plots (a) and ratios (b) of different groups of T cells in the residual tumours from different groups (gated on CD3⁺ T cells, n = 5). Cells were stained with anti‐CD3‐FITC, anti‐CD4‐Brilliant Violet 421, anti‐CD8a‐APC/Fire750 antibodies (Biolegend). Error bar, mean ± s.d.. (c, d) Representative flow cytometry plots (c) and ratios (d) of different groups of CD8⁺ Ki67⁺ T cells in tumours (gated on CD3⁺ CD8⁺ T cells, n = 5). Cells were stained with anti‐CD3‐FITC, anti‐CD8‐APC/Fire750, anti‐Ki67‐Alexa Fluor^® 647 antibodies (Biolegend). Error bar, mean ± s.d.. (e, f) Representative flow cytometry plots (e) and ratios (f) of different groups of CD8⁺ CD69⁺ T cells in tumours (Gated on CD3⁺ CD8⁺ T cells, n = 5). Cells were stained with anti‐CD3‐FITC, anti‐CD8‐APC/Fire750, anti‐CD69‐APC antibodies (Biolegend). Error bar, mean ± s.d.. (g, h) Representative flow cytometry plots (g) and ratios (h) of different groups of CD44⁺ CD62L⁺ T cells infiltrating in tumours (Gated on CD3⁺ CD8⁺ T cells, n = 5). Cells were stained with anti‐CD3‐FITC, anti‐CD8‐APC/Fire750, anti‐CD44‐PE, anti‐CD62L‐Alexa Fluor^® 700 antibodies (Biolegend). Error bar, mean ± s.d.. (i) The immunofluorescence images of CD4⁺ CD8⁺ T cells infiltrating in tumours (Scale bar: 50 μm). (#1) Gel‐PBS, (#2) Gel‐Blank BDVs, (#3) Gel‐Normal‐M33‐M47 BDVs, (#4) Gel‐Mutation‐M33‐M47 BDVs, (#5) aPD‐1, (#6) Gel‐Mutation‐M33‐M47 BDVs + aPD‐1. Cells were stained with anti‐CD11c‐APC, anti‐CD40‐PE, anti‐CD80‐FITC and anti‐CD86‐PE antibodies (Biolegend). All samples were sorted using Beckman CytoFlex flow cytometer and analyzed with FlowJo software. NS: no significant, *P < 0.05, **P < 0.01, ***P < 0.001. One‐way ANOVA with Tukey post‐hoc tests (b, d, f, h)

FIGURE 5

In vivo suppression of antimetastatic tumour effect of BDV‐neoantigen vaccine. (a) Schematic illustration of the therapy of BDV‐neoantigen vaccine in B16F10‐luc model of metastasis. (b) In vivo bioluminescence images of the B16F10‐luc lung metastasis in different groups (n = 5). (c) Representative images of the lung metastatic nodules (Scale bar: 5 mm). (d) Numbers of lung metastatic nodules in different groups (n = 5), Error bar, mean ± s.d.. (e) Survival curves of different groups (n = 10). Day 0 is the day of tumour cells injection via tail vein. (f) H&E‐stained lung slices (Scale bar: 100 μm). (#1) Gel‐PBS, (#2) Gel‐Blank BDVs, (#3) Gel‐Normal‐M33‐M47 BDVs, (#4) Gel‐Mutation‐M33‐M47 BDVs, (#5) aPD‐1, (#6) Gel‐Mutation‐M33‐M47 BDVs + aPD‐1. NS: no significant, *P < 0.05, **P < 0.01, ***P < 0.001. One‐way ANOVA with Tukey post‐hoc tests (d) or the Long‐Rank (Mantel‐Cox) test (e)